Hire the Best Recommender Systems Specialists

More than 3,000 reviews on G2
Rating is 4.5 out of 5.
4.5/5
of Upwork by G2 peer reviewers
Haithem G.

Setif, Algeria

$45/hr
4.9
104 jobs

๐Ÿ‘‹ Hi! I'm Haithem, and I love to help businesses shine with custom AI applications, workflow automation, AI Agent integrations and intelligent chatbots. I leverage the latest AI advancements and modern automation tools to craft tailored solutions that are practical, efficient, and ready to use. My mission is to bridge the gap between cutting-edge technology and everyday business needs, ensuring you get solutions that not only look smart but also work smart. Here's a snapshot of my expertise: โœ” Mastery in deploying AI agents for dynamic web and API applications, enabling businesses to streamline operations and enhance decision-making processes. โœ” Extensive experience designing agentic systems with Claude Code, Claude Opus/Sonnet/Haiku, Gemini and GPT-5.5; including long-running agents, sub-agents, Skills, and reflection/critique loops. โœ” Specialist in AI integrations, seamlessly incorporating AI functionalities into existing platforms and workflows, from automating repetitive tasks to enriching customer interactions. โœ” Automation expertise using tools like Make .com, n8n, Zapier, Gumloop, Lindy, Relevance AI, CrewAI, LangChain / LangGraph & GoHighLevel (GHL). โœ” Proficient in building and integrating custom AI voice agents with VAPI, Retell AI, ElevenLabs in addition to AI-powered chatbots for a range of use cases, including customer support, lead generation, cold calling, customer outreach and internal process optimization. โœ” Skilled at designing end-to-end solutions combining Airtable and Supabase databases, Make scenarios, and API-driven workflows for no-code/low-code development. โœ” Specialist in building custom MCP (Model Context Protocol) servers that expose internal APIs, databases, and SaaS tools to Claude, Cursor, and Claude Code. โœ” Proficient in RAG and agentic RAG pipelines over Pinecone, Weaviate, Qdrant, and Supabase pgvector; with citations, guardrails, and hallucination controls. โœ” Deep CRM and stack integrations across GoHighLevel, HubSpot, Salesforce, Pipedrive, Close .com, Monday .com, Airtable, Notion, ServiceTitan, Housecall Pro, Jobber and many other 3rd party tools. โœ” Expertise in prompt engineering, with a proven track record of formulating prompts that deliver contextually aware AI responses. โœ” AI video, avatar, and character generation with HeyGen, Synthesia, Tavus, Runway, Pika, Sora, Luma AI, Kling, Hailuo, Higgsfield AI, and Captions AI; for UGC ads, talking head explainers, personalized outreach videos, course content, and multilingual brand spokespersons at scale. ๐Ÿค– Here are some of the Automation Workflows and AI Agents I worked on using Make, VAPI and other automation tools: โœ… AI Voice Agents: for dental and medical clinics, real estate agencies, restaurants, banking and financial services, marketing agencies, insurance companies, hotels and accomodations, private schools, travel agencies and more. โœ… AI cold caller using Make, Retell or VAPI and Airtable. โœ… Make AI Workflow for Social Media Content Creation, Publishing and Distribution. โœ… Data Scraping and Lead Generation Automation System using Make and Apify. โœ… Automate AI Video Content Creation For Any Platform (Reels, Shorts, TikTok) using Make. Below are some of the AI-powered solutions I've worked on before: โœ” ๐€๐ˆ ๐‘๐ž๐œ๐ž๐ฉ๐ญ๐ข๐จ๐ง๐ข๐ฌ๐ญ ๐Ž๐’: A multi-tenant voice receptionist platform for home-service businesses (HVAC, plumbing, dental, auto repair) that answers every call 24/7, qualifies the caller, books the job on Cal .com or ServiceTitan, dispatches techs via SMS, and logs the full transcript and outcome to GHL or HubSpot. Built on Retell AI with ElevenLabs voice cloning and Twilio SIP trunking. โœ” ๐€๐ˆ ๐Ž๐ฎ๐ญ๐›๐จ๐ฎ๐ง๐ ๐’๐š๐ฅ๐ž๐ฌ ๐„๐ง๐ ๐ข๐ง๐ž: A full agentic SDR stack that scrapes Apollo, Phantombuster, and Hunter, enriches every lead with Clay, scores fit with Claude Opus, drafts and sends hyper-personalized cold emails and LinkedIn DMs through Smartlead and Instantly, then classifies replies and books qualified meetings straight into the rep's calendar; deployed for B2B SaaS and agency clients running 5K+ leads per week. โœ” ๐€๐ˆ ๐Ž๐ฉ๐ž๐ซ๐š๐ญ๐ข๐จ๐ง๐ฌ ๐ƒ๐š๐ฌ๐ก๐›๐จ๐š๐ซ๐: A daily executive briefing system for multi-location salon, spa, and service businesses that pulls KPIs from Phorest, Xero, HubSpot, and ServiceTitan, runs them through GPT-4o for anomaly detection and natural-language commentary, and emails owners a personalized morning dashboard with red flags, wins, and recommended actions; built on Make, Airtable, and a Claude critique pass that keeps the narrative honest. โœ” ๐€๐ˆ ๐€๐ฏ๐š๐ญ๐š๐ซ ๐•๐ข๐๐ž๐จ ๐„๐ง๐ ๐ข๐ง๐ž: An automated UGC video production system that turns a single brand brief into hundreds of personalized videos at scale using HeyGen, with auto-captioning and publishing across TikTok, Reels, Shorts, YouTube, and email outreach. Letโ€™s work together to bring innovative, intelligent, automated and AI-based solutions to your business!๐Ÿš€

  • Natural Language Processing
  • AI App Development
  • Retrieval Augmented Generation
  • Large Language Model
  • Artificial Intelligence
  • Chatbot Development
  • AI Agent Development
  • Automation
  • Make.com
  • AI Development
  • HighLevel
  • API Integration
  • n8n
  • Claude
  • Marketing Automation
  • Zapier
  • Conversational AI
  • Automated Workflow
  • Airtable
  • Business Process Automation
Francisco G.

Boston, Massachusetts

$150/hr
5.0
2 jobs

I've built AI that understands millions of ๐€๐ฅ๐ž๐ฑ๐š requests, ranks what millions of ๐‘๐จ๐ค๐ฎ viewers watch next, and personalizes what shoppers see at fast-growing ๐ž-๐œ๐จ๐ฆ๐ฆ๐ž๐ซ๐œ๐ž brands. Now I bring that same caliber of ๐ฆ๐š๐œ๐ก๐ข๐ง๐ž ๐ฅ๐ž๐š๐ซ๐ง๐ข๐ง๐  to your product. For ๐Ÿ๐ŸŽ+ ๐ฒ๐ž๐š๐ซ๐ฌ โ€” ๐๐ก๐ƒ in machine learning, ๐๐ž๐ฎ๐ซ๐ˆ๐๐’ & ๐๐€๐€๐‚๐‹ published, patented in ๐๐‹๐ โ€” I've shipped ๐ฉ๐ซ๐จ๐๐ฎ๐œ๐ญ๐ข๐จ๐ง ๐€๐ˆ at some of the biggest tech companies in the world. ๐“๐จ๐๐š๐ฒ I bring that experience directly to businesses of any size, from your first AI feature to systems serving millions of users. WHAT I'VE SHIPPED - Candidate generation and ranking for the recommendation system of a top shopping app used by millions of shoppers, including LLM-based recommendation techniques - A store-wide recommendation engine โ€” and the company personalization roadmap behind it โ€” for a fast-growing e-commerce brand - Recommendation and ranking models served to millions of streaming households at Roku - Core natural language understanding models behind Amazon Alexa โ€” the systems that decide what you meant and what should happen next - Peer-reviewed AI research at NeurIPS, NAACL, and ICRA, plus a granted NLP patent WHAT I CAN BUILD FOR YOU - Recommendation systems & personalization: "customers also bought," homepage and email personalization, search ranking, content feeds โ€” candidate generation, ranking, contextual bandits, and real-time personalization that lift conversion, engagement, and retention - LLM applications & AI agents: AI agents, RAG pipelines over your docs or product catalog, semantic search, fine-tuning, and LLM feature integration โ€” the retrieval and ranking architecture I've shipped for a decade, now paired with LLMs in production - AI chatbots & conversational AI: support bots, product Q&A, intent understanding, and entity extraction โ€” grounded in your data so they answer correctly - Data science & predictive modeling: demand forecasting, customer segmentation, churn prediction, and A/B testing you can act on - Research to production: translating papers or early-stage ideas into working systems โ€” I'm at my best when the problem isn't fully defined yet and the objective still needs clarifying - End-to-end ML engineering: architecture reviews, model development, deployment, MLOps, and high-performance inference โ€” systems that survive real traffic, not notebooks WHO I WORK WITH E-commerce and DTC brands, marketplaces, streaming and media platforms, and SaaS teams โ€” whether you're adding your first AI feature or scaling an ML system that's hitting its limits. TOOLS & SKILLS - ML & modeling: Python, PyTorch, deep learning, reinforcement learning, contextual bandits, ranking models - LLM & generative AI: GPT, Claude, Gemini and open-source models, AI agents, RAG, embeddings & vector databases, semantic search, fine-tuning - Data & infrastructure: SQL, Snowflake, Snowpark, PySpark, AWS, MLOps, high-performance inference - Measurement: A/B testing, experiment design, offline & online evaluation, recommendation metrics (CTR, conversion, retention) HOW WE CAN ENGAGE - Strategy consultation (60 min): straight answers on your AI roadmap and what to build first โ€” before you spend on development - Fixed-scope projects: architecture reviews, audits, and MVPs with a defined deliverable, timeline, and price - Ongoing development: hourly engagement to build, scale, and maintain your ML and AI systems WHY CLIENTS PICK ME - 10+ years of production machine learning at Amazon, Roku, and fast-growing e-commerce brands - PhD in machine learning (UMass Amherst); first-author publications at NeurIPS and NAACL - Reviewer for the world's top AI conferences: NeurIPS, ICML, ICLR, AAAI - I've owned AI strategy, not just code โ€” roadmaps and priorities as well as models - Fluent English, native Spanish โ€” I work seamlessly with US, European, and Latin American teams HOW I WORK Every project starts with clear scoping and an honest estimate: what we're building, what "working" means, and how we'll measure it. Then I build in milestones you can see โ€” no black boxes, no jargon walls. You get production-quality code, documentation, and plain-language explanations of every decision. And I'll tell you when a simpler model beats the fancy one โ€” or when AI isn't the answer at all. That honesty is cheaper on day one than on day ninety. Message me with a few lines about your product and what you're trying to improve. I'll give you a straight answer on what will work, what it takes, and what it's worth.

  • Machine Learning
  • Artificial Intelligence
  • Generative AI
  • Reinforcement Learning
  • Amazon Bedrock
  • Amazon SageMaker
  • Retrieval Augmented Generation
  • Large Language Model
  • AI Agent Development
  • Natural Language Processing
  • CUDA
  • Shopify
  • GPU
  • Recommendation System
  • A/B Testing
  • Python
  • PyTorch
  • Software Architecture
  • Vector Database
  • Chatbot
Ratnapriya L.

Ahmedabad, India

$25/hr
4.4
3 jobs

Hi! Iโ€™m Ratnapriya, a Senior Data Scientist specializing in Machine Learning, NLP and predictive modeling. I build end-to-end solutions that turn audio, text and tabular data into reliable, production-ready systems; from data design and feature engineering to model training, evaluation and deployment. What I deliver: โ€ข Text summarization, sentiment, topic & NLP pipelines (transformers + classical) โ€ข Speech emotion & audio analysis (MFCCs, spectrograms) โ€ข Predictive models & lead scoring (XGBoost, tabular pipelines) โ€ข Clustering/EDA for root-cause analysis and automation workflows I focus on clear results, reproducible code, and outputs non-technical teams can use. I can start immediately.

  • Machine Learning
  • Data Science
  • Natural Language Processing
  • Predictive Modeling
  • Python
  • XGBoost
  • Hugging Face
  • TensorFlow
  • PyTorch
  • Data Analysis
  • Python Scikit-Learn
  • Speech Emotion Recognition
  • PostgreSQL
  • Feature Engineering
  • Unsupervised Learning
Ali A.

Lahore, Pakistan

$30/hr
5.0
16 jobs

โœ… ๐’๐ž๐ง๐ข๐จ๐ซ ๐€๐ˆ ๐„๐ง๐ ๐ข๐ง๐ž๐ž๐ซ | ๐Ÿ๐ŸŽ+ ๐˜๐ž๐š๐ซ๐ฌ | ๐‹๐‹๐Œ๐ฌ, ๐‘๐€๐†, ๐‚๐จ๐ฆ๐ฉ๐ฎ๐ญ๐ž๐ซ ๐•๐ข๐ฌ๐ข๐จ๐ง & ๐€๐ˆ ๐€๐ฎ๐ญ๐จ๐ฆ๐š๐ญ๐ข๐จ๐ง If you are looking for an AI engineer who can take your idea beyond a proof of concept and turn it into a reliable production system, you are in the right place. โœ” 10+ Years of AI and Machine Learning Experience โœ” Production Ready AI Systems Across Multiple Industries โœ” LLMs, AI Agents, Computer Vision, RAG and Enterprise Automation I help startups, enterprises, and growing businesses design, build, and deploy intelligent AI solutions that automate work, extract insights from data, improve decision making, and integrate seamlessly into existing products and workflows. My experience spans healthcare, oil and gas, fintech, ecommerce, utilities, industrial automation, and enterprise software. Alongside delivering commercial AI solutions, my research work has been associated with leading institutions including MIT, Georgia Tech, Brown University, and the University of Michigan, giving me the ability to combine cutting edge research with practical engineering that delivers measurable business value. ๐€๐ˆ ๐’๐จ๐ฅ๐ฎ๐ญ๐ข๐จ๐ง๐ฌ ๐ˆ ๐๐ฎ๐ข๐ฅ๐ ๐‘ณ๐’‚๐’“๐’ˆ๐’† ๐‘ณ๐’‚๐’๐’ˆ๐’–๐’‚๐’ˆ๐’† ๐‘ด๐’๐’…๐’†๐’๐’” ๐’‚๐’๐’… ๐‘จ๐‘ฐ ๐‘จ๐’ˆ๐’†๐’๐’•๐’” Build intelligent assistants, AI copilots, multi agent systems, workflow automation, and business process automation using OpenAI, Claude, Gemini, LangGraph, CrewAI, AutoGen, Model Context Protocol, and custom backend services. ๐‘น๐’†๐’•๐’“๐’Š๐’†๐’—๐’‚๐’ ๐‘จ๐’–๐’ˆ๐’Ž๐’†๐’๐’•๐’†๐’… ๐‘ฎ๐’†๐’๐’†๐’“๐’‚๐’•๐’Š๐’๐’ ๐’‚๐’๐’… ๐‘ฒ๐’๐’๐’˜๐’๐’†๐’…๐’ˆ๐’† ๐‘บ๐’š๐’”๐’•๐’†๐’Ž๐’” Develop private AI assistants that securely work with your documents and company knowledge using vector databases, semantic search, document indexing, and enterprise search solutions powered by Pinecone, Weaviate, ChromaDB, FAISS, and modern LLM frameworks. ๐‘ช๐’๐’Ž๐’‘๐’–๐’•๐’†๐’“ ๐‘ฝ๐’Š๐’”๐’Š๐’๐’ ๐’‚๐’๐’… ๐‘ซ๐’†๐’†๐’‘ ๐‘ณ๐’†๐’‚๐’“๐’๐’Š๐’๐’ˆ Design custom computer vision solutions for object detection, segmentation, defect detection, medical imaging, industrial inspection, video analytics, tracking, OCR, and image understanding using PyTorch, TensorFlow, YOLO, SAM, and advanced deep learning architectures. ๐‘ฌ๐’๐’… ๐’•๐’ ๐‘ฌ๐’๐’… ๐‘จ๐‘ฐ ๐‘ท๐’“๐’๐’…๐’–๐’„๐’•๐’” Build complete AI powered applications from backend APIs to deployment using Python, FastAPI, Flask, Docker, cloud infrastructure, model serving, monitoring, and scalable production pipelines. ๐–๐ก๐š๐ญ ๐ˆ ๐‚๐š๐ง ๐‡๐ž๐ฅ๐ฉ ๐˜๐จ๐ฎ ๐๐ฎ๐ข๐ฅ๐ โ€ข AI Agents for business automation โ€ข Enterprise knowledge assistants โ€ข Document processing and intelligent data extraction โ€ข Chatbots powered by private company data โ€ข Text to SQL and business intelligence systems โ€ข Computer vision inspection platforms โ€ข Medical imaging and healthcare AI โ€ข Predictive analytics and forecasting โ€ข Recommendation systems โ€ข AI powered SaaS products โ€ข Custom Machine Learning models โ€ข End to end AI platforms ๐’๐ž๐ฅ๐ž๐œ๐ญ๐ž๐ ๐๐ซ๐จ๐ฃ๐ž๐œ๐ญ ๐„๐ฑ๐ฉ๐ž๐ซ๐ข๐ž๐ง๐œ๐ž โ€ข Built an enterprise RAG assistant enabling employees to search thousands of internal documents using natural language. โ€ข Developed AI powered document extraction systems that transformed unstructured documents into structured business data. โ€ข Created a Text to SQL platform allowing business users to query databases without writing SQL. โ€ข Delivered computer vision systems for crack detection, debris detection, and industrial infrastructure monitoring. โ€ข Built logo detection models using Mask R CNN for sports analytics. โ€ข Developed depth estimation and three dimensional reconstruction pipelines from image sequences. โ€ข Fine tuned generative AI models for custom image generation and synthetic content creation. ๐–๐ก๐ฒ ๐‚๐ฅ๐ข๐ž๐ง๐ญ๐ฌ ๐‚๐ก๐จ๐จ๐ฌ๐ž ๐Œ๐ž I focus on building AI systems that are reliable, scalable, and ready for production rather than experimental prototypes. When we work together, you can expect: โœ” Clear communication throughout the project โœ” Clean, maintainable, and scalable code โœ” Practical solutions focused on business outcomes โœ” Strong software engineering and AI architecture โœ” Reliable deployment and long term maintainability Whether you need an AI automation workflow, a private knowledge assistant, a computer vision solution, or a complete AI powered software product, I can help you design, build, deploy, and scale it with confidence. ๐ฟ๐‘’๐‘ก'๐‘  ๐‘‘๐‘–๐‘ ๐‘๐‘ข๐‘ ๐‘  ๐‘ฆ๐‘œ๐‘ข๐‘Ÿ ๐‘๐‘Ÿ๐‘œ๐‘—๐‘’๐‘๐‘ก ๐‘Ž๐‘›๐‘‘ ๐‘ก๐‘ข๐‘Ÿ๐‘› ๐‘ฆ๐‘œ๐‘ข๐‘Ÿ ๐ด๐ผ ๐‘ฃ๐‘–๐‘ ๐‘–๐‘œ๐‘› ๐‘–๐‘›๐‘ก๐‘œ ๐‘Ž ๐‘๐‘Ÿ๐‘œ๐‘‘๐‘ข๐‘๐‘ก๐‘–๐‘œ๐‘› ๐‘Ÿ๐‘’๐‘Ž๐‘‘๐‘ฆ ๐‘ ๐‘œ๐‘™๐‘ข๐‘ก๐‘–๐‘œ๐‘›.

  • Python
  • Computer Vision
  • Data Science
  • TensorFlow
  • Deep Learning
  • Machine Learning
  • Natural Language Processing
  • Reinforcement Learning
  • PyTorch
  • Data Science Consultation
  • Artificial Neural Network
  • Artificial Intelligence
  • Embedded System
  • Mathematical Optimization
  • Mathematical Modeling
Adam M.

Manchester, United Kingdom

$125/hr
5.0
88 jobs

I build production AI systems that businesses run on every day: AI agents, RAG pipelines and LLM workflow automation for companies where a wrong answer costs real money. WHY CLIENTS PICK ME โ€ข Expert-Vetted, the badge Upwork awards its top 1% by interview, not by algorithm. 100% Job Success across 70+ projects and six years. โ€ข Production, not prototypes. Not proof-of-concepts. Every system ships with schema validation, evals and a regression suite, so it still works six months after launch and you can prove it. โ€ข Scale. I have processed 12 million documents through a single RAG pipeline and run agents live in front of real consumers. โ€ข I integrate with what you already have. Your CRM, your database, your APIs, your cloud. AWS certified (ML Specialty), equally at home on GCP. โ€ข You always know where the project is. A written update whenever something moves and a weekly Loom walkthrough. Clients tell me this is the part they remember. WHAT HAPPENS WHEN THIS GOES WELL A process that eats your team's week runs on its own, and you get the evidence it did not get worse. On an AI underwriting platform used by 15+ insurers across the US, policy review went from days to 3 to 5 minutes at 99.55% precision on the rules that move money. A UK mortgage broker cut manual payslip review by 80%, because the automation only escalates the documents two models disagree on. A UK government-data client got 12 million planning documents extracted, embedded and searchable in under 48 hours at 65% below the GPT-4 baseline cost. Most of my engagements run long-term. The first project automates one workflow; the ones after that tend to automate the rest. WHAT I DO 1. AI agents and workflow automation. Multi-agent systems that take a manual process end to end and make real decisions along the way. Built in LangGraph and Claude with MCP, schema-validated, observable at every step, with a human in the right place by design. I built the assistant inside a UK consumer money app: seven specialised agents, 43 deterministic tools, every financial figure computed by code rather than the model, so it cannot make up a number. 2. RAG and LLM engineering. Retrieval pipelines, knowledge assistants and chatbots over your own data, and cost and accuracy work on LLM systems already in production. 700+ UK planning policy documents summarised weekly, fully automated, at 75% lower LLM cost through context caching and batch APIs. 3. The machine learning and data foundations underneath. Document and data extraction, classification, Python, SQL, pandas and scikit-learn. A private-equity deal-screening agent I built completes the firm's own investment scorecard end to end, backtested against realised fund returns. HOW I BUILD The simplest thing that works, then iterate with evidence. Strong prompting before fine-tuning. An API call before custom infrastructure. A benchmark before an architecture decision. Non-negotiable on every build: โ€ข Structured outputs with schema enforcement. If it does not validate, it does not pass โ€ข Dual-model verification on high-stakes data โ€ข Full observability with LangSmith or Langfuse. Nothing is a black box โ€ข A gold-standard eval set built early and regression-tested on every change WHAT CLIENTS SAY "His Loom updates were the highlight of my week! He's great at communicating complex AI concepts to non-technical people and keeps you in the loop every step of the way." Chris Barnes, Co-Founder, Gains App "Adam is an absolute powerhouse of an LLM Engineer. He has first class communication skills which make working with him an absolute pleasure." Sammie Ellard-King, Founder, Gains App "A great balance of personality, professionalism and a deep knowledge of the AI space. He's a strategic thinker who considers the bigger picture." Tom Story, PlannrAI STACK โ€ข Agents and orchestration: LangGraph, LangChain, Claude Code, MCP, LangSmith, Langfuse โ€ข Models: Claude, OpenAI, ChatGPT, Gemini, Llama, via Bedrock, Vertex and direct APIs โ€ข Backend and infrastructure: Python, FastAPI, Postgres with pgvector, Docker, Kubernetes, AWS (ML Specialty certified), GCP โ€ข Machine learning: PyTorch, scikit-learn, pandas, SQL โ€ข Reliability: Pydantic structured outputs, dual-LLM verification, evals and LLM-as-judge test suites TAKING ON NOW โ€ข AI agent and workflow automation builds, end to end โ€ข RAG pipelines and knowledge assistants over your own data โ€ข Cost and accuracy work on LLM systems already in production โ€ข Claude Code and AI engineering enablement for teams adopting AI Expert-Vetted ยท 100% Job Success ยท $500K+ earned ยท 5,850+ hours ยท 75+ projects

  • Python
  • Deep Learning
  • PyTorch
  • Data Analysis
  • Machine Learning
  • Data Science
  • Artificial Intelligence
  • AI Agent Development
  • LangChain
  • Retrieval Augmented Generation
  • Large Language Model
  • Prompt Engineering
  • AI Consulting
  • Generative AI
  • AI Chatbot
  • Automation
  • Amazon Web Services
  • API Integration
  • ChatGPT
  • Data Extraction
Uzma A.

Karachi, Pakistan

$33/hr
5.0
2 jobs

๐—œ ๐—ฏ๐˜‚๐—ถ๐—น๐—ฑ ๐—”๐—œ ๐˜€๐˜†๐˜€๐˜๐—ฒ๐—บ๐˜€ ๐˜๐—ต๐—ฎ๐˜ ๐˜€๐—ฎ๐˜ƒ๐—ฒ ๐—ฏ๐˜‚๐˜€๐—ถ๐—ป๐—ฒ๐˜€๐˜€๐—ฒ๐˜€ $30๐—žโ€“$40๐—ž ๐˜†๐—ฒ๐—ฎ๐—ฟ๐—น๐˜† ๐—ฏ๐˜† ๐—ฎ๐˜‚๐˜๐—ผ๐—บ๐—ฎ๐˜๐—ถ๐—ป๐—ด ๐—ผ๐—ฝ๐—ฒ๐—ฟ๐—ฎ๐˜๐—ถ๐—ผ๐—ป๐˜€, ๐˜€๐—ฎ๐—น๐—ฒ๐˜€, ๐—ฎ๐—ป๐—ฑ ๐˜€๐˜‚๐—ฝ๐—ฝ๐—ผ๐—ฟ๐˜ ๐—”๐—œ ๐—ฆ๐—ผ๐—น๐˜‚๐˜๐—ถ๐—ผ๐—ป๐˜€ ๐—ณ๐—ผ๐—ฟ ๐—•๐˜‚๐˜€๐—ถ๐—ป๐—ฒ๐˜€๐˜€๐—ฒ๐˜€ | ๐Ÿญ๐˜… ๐—ž๐—ฎ๐—ด๐—ด๐—น๐—ฒ ๐—š๐—ฟ๐—ฎ๐—ป๐—ฑ๐— ๐—ฎ๐˜€๐˜๐—ฒ๐—ฟ | ๐Ÿญ๐˜… ๐—ž๐—ฎ๐—ด๐—ด๐—น๐—ฒ ๐—ก๐—ผ๐˜๐—ฒ๐—ฏ๐—ผ๐—ผ๐—ธ ๐—˜๐˜…๐—ฝ๐—ฒ๐—ฟ๐˜ | ๐— ๐—ฎ๐—ฐ๐—ต๐—ถ๐—ป๐—ฒ ๐—Ÿ๐—ฒ๐—ฎ๐—ฟ๐—ป๐—ถ๐—ป๐—ด | ๐—”๐—œ ๐—˜๐—ป๐—ด๐—ถ๐—ป๐—ฒ๐—ฒ๐—ฟ | ๐—”๐—œ ๐—”๐—ฝ๐—ฝ ๐——๐—ฒ๐˜ƒ๐—ฒ๐—น๐—ผ๐—ฝ๐—ฒ๐—ฟ | ๐—ฅ๐—”๐—š ๐—ฆ๐˜†๐˜€๐˜๐—ฒ๐—บ๐˜€ (๐—Ÿ๐—ฎ๐—ป๐—ด๐—–๐—ต๐—ฎ๐—ถ๐—ป, ๐—Ÿ๐—ฎ๐—ป๐—ด๐—š๐—ฟ๐—ฎ๐—ฝ๐—ต) | ๐—ฃ๐˜†๐˜๐—ต๐—ผ๐—ป | ๐—™๐—ฎ๐˜€๐˜๐—”๐—ฃ๐—œ ๐—ง๐—ฒ๐—ฐ๐—ต๐—ป๐—ถ๐—ฐ๐—ฎ๐—น ๐—–๐—ผ๐—ป๐˜๐—ฒ๐—ป๐˜ ๐—–๐—ฟ๐—ฒ๐—ฎ๐˜๐—ผ๐—ฟ I'm a Senior ML Engineer and Kaggle GrandMaster who builds production-grade AI systems from data pipelines to deployed APIs. I help businesses turn messy data and AI ideas into reliable, working products, across the full AI stack: machine learning, generative AI, computer vision, and data engineering. ๐—Ÿ๐—Ÿ๐— , ๐—ฅ๐—”๐—š & ๐—”๐—œ ๐—”๐—ด๐—ฒ๐—ป๐˜๐˜€: LLM integration, RAG pipelines, AI agent development, multi-agent systems, prompt engineering, fine-tuning, vector databases (Pinecone, ChromaDB, FAISS), semantic search, embeddings, AI chatbots and copilots built with LangChain, LangGraph, OpenAI API, and Hugging Face Transformers. ๐— ๐—ฎ๐—ฐ๐—ต๐—ถ๐—ป๐—ฒ ๐—Ÿ๐—ฒ๐—ฎ๐—ฟ๐—ป๐—ถ๐—ป๐—ด & ๐——๐—ฎ๐˜๐—ฎ ๐—ฆ๐—ฐ๐—ถ๐—ฒ๐—ป๐—ฐ๐—ฒ: Predictive modeling, classification, regression, time series forecasting, clustering, feature engineering, model evaluation, A/B testing, anomaly detection, and statistical analysis turning raw data into measurable business outcomes like reduced error rates and faster decisions. ๐—–๐—ผ๐—บ๐—ฝ๐˜‚๐˜๐—ฒ๐—ฟ ๐—ฉ๐—ถ๐˜€๐—ถ๐—ผ๐—ป & ๐—š๐—ฒ๐—ป๐—ฒ๐—ฟ๐—ฎ๐˜๐—ถ๐˜ƒ๐—ฒ ๐—”๐—œ: Image classification, object detection, image segmentation, OCR, image generation (Stable Diffusion, GANs), and deep learning model training using PyTorch and TensorFlow. ๐—”๐—œ ๐—˜๐—ป๐—ด๐—ถ๐—ป๐—ฒ๐—ฒ๐—ฟ๐—ถ๐—ป๐—ด & ๐——๐—ฒ๐—ฝ๐—น๐—ผ๐˜†๐—บ๐—ฒ๐—ป๐˜: Model deployment, MLOps, REST API development (FastAPI), containerization (Docker), cloud deployment (AWS), CI/CD for ML, scalable backend systems for AI apps, and end-to-end pipeline automation. ๐——๐—ฎ๐˜๐—ฎ ๐—˜๐—ป๐—ด๐—ถ๐—ป๐—ฒ๐—ฒ๐—ฟ๐—ถ๐—ป๐—ด & ๐—”๐—ป๐—ฎ๐—น๐˜†๐˜๐—ถ๐—ฐ๐˜€: ETL pipelines, data cleaning, feature stores, SQL, dashboarding and reporting (Streamlit, Tableau), and exploratory data analysis. ๐—ง๐—ฒ๐—ฐ๐—ต ๐˜€๐˜๐—ฎ๐—ฐ๐—ธ: Python, PyTorch, TensorFlow, scikit-learn, Pandas, NumPy, Hugging Face, LangChain, LangGraph, OpenAI API, SQL, FastAPI, Docker, AWS, Streamlit, Tableau, Pinecone, ChromaDB Kaggle Grand Master verify my rank: ๐—ธ๐—ฎ๐—ด๐—ด๐—น๐—ฒ: ๐˜‚๐˜‡๐—บ๐—ฎ๐—ฎ๐—ธ๐—ต๐˜๐—ฎ๐—ฟ ๐—ข๐—ฝ๐—ฒ๐—ป-๐˜€๐—ผ๐˜‚๐—ฟ๐—ฐ๐—ฒ ๐˜„๐—ผ๐—ฟ๐—ธ: Github As a Technical Content Creator, I also document and share my AI/ML work publicly so you can review real code, notebooks, and results before you hire, not just promises. ๐—–๐—ผ๐—ฟ๐—ฒ ๐—ฝ๐—ฟ๐—ถ๐—ป๐—ฐ๐—ถ๐—ฝ๐—น๐—ฒ๐˜€: focus, consistency, attention to detail, and reliable delivery. Let's talk about what you're trying to build.

  • Artificial Intelligence
  • Machine Learning Model
  • Large Language Model
  • LangChain
  • Retrieval Augmented Generation
  • Natural Language Processing
  • Data Science
  • Python
  • PyTorch
  • TensorFlow
  • FastAPI
  • AWS Glue
  • Data Analysis
  • AI Chatbot
  • AI Agent Development
  • Deep Learning
  • MLOps
  • SQL

How it works

Post a job for freePost a job

Tell us what you need. Create your own job post or generate one with AI then filter talent matches.

Hire top talent fast

Consult, interview, and hire quickly, so you can meet the freelancers you're excited about.

Collaborate easily

Use Upwork to chat or video call, share files, and track project progress right from the app.

Payment simplified

Manage payments in one place with flexible billing options. Only pay for approved work, hourly or by milestone.

Don't just take our word for it

What does a Recommender systems specialist do?

A recommender systems specialist builds machine learning pipelines that rank items for specific users based on their past behavior and preferences. This role focuses on transforming raw interaction data into personalized lists that appear in search results, product feeds, or content streams. You design the logic that decides which items a user sees first, balancing relevance with business goals like diversity or freshness. Your work directly influences user engagement by surfacing the most useful options from a large catalog.

  • You analyze user interaction logs and item metadata to define clear recommendation objectives such as click-through rate improvement or conversion growth. This step involves cleaning data sets to remove noise and selecting features that accurately represent user intent and item characteristics. You establish baseline metrics to measure how well current systems perform before introducing new models.
  • You develop multi-stage recommender pipelines that include candidate generation, scoring, and re-ranking components to handle large-scale data efficiently. Candidate generation retrieves a broad set of potential matches using retrieval methods like embedding-based search or collaborative filtering. Scoring models then predict the likelihood of user interest for each candidate, while re-ranking applies business rules to adjust the final order. You write code that connects these stages into a cohesive workflow capable of processing millions of interactions.
  • You train and validate models using offline evaluation frameworks to compare different algorithms against ranking metrics such as precision at k or normalized discounted cumulative gain. This process requires running experiments to test how changes in model architecture or feature engineering affect prediction accuracy. You generate detailed reports that document model performance differences and justify the selection of specific approaches for production use. Your analysis helps the team avoid deploying models that might degrade the user experience.
  • You collaborate with software engineers to deploy recommendation models into production environments where they serve real-time requests from users. This work includes optimizing inference latency to ensure recommendations load quickly without slowing down the application. You monitor system performance after deployment to detect drift in data patterns or drops in recommendation quality. You update models regularly to incorporate new user behavior and maintain high relevance over time.

How to hire a Recommender systems specialist on Upwork

Step 1: Post a job

Define your recommendation objectives and data constraints clearly to attract qualified candidates. Use the Job Post Generator powered by Umaโ„ข, Upwork's Mindful AI to draft a precise description from a few sentences about your needs. You can write a new post, update a saved draft, or reuse an existing post.

  • Specify whether you need candidate generation, scoring, or re-ranking pipelines built for your specific user base.
  • List the machine learning frameworks and data processing infrastructure your team currently uses for model training.
  • Detail the offline evaluation metrics and ranking quality standards the specialist must meet during validation.

Step 2: Evaluate candidates

Look for portfolios that demonstrate end-to-end system performance improvements in production environments. Uma can run instant video interviews and build shortlists with side-by-side comparisons to help you assess technical depth quickly.

  • Review code samples that show how the candidate preprocesses interaction data and defines feature sets for users and items.
  • Check for documented offline evaluation results that compare different policies using standard ranking-based metrics.
  • Verify experience deploying recommendation models and integrating them into live search or retrieval components.

Step 3: Interview your top choices

Discuss their approach to balancing accuracy with computational constraints in real-time serving scenarios. Schedule and conduct these interviews within Upwork Messages to receive an immediate transcript and summary after each session.

  • Ask how they optimize embedding-based representations for faster candidate generation without losing personalization quality.
  • Request examples of how they collaborated with product teams to translate business goals into ranking objectives.
  • Explore their method for monitoring model drift and updating training pipelines when user behavior changes.

Step 4: Agree on scope and begin work

Set clear milestones for model development, offline testing, and production deployment artifacts. Use Upwork Messages and the contract workroom for communication and project management, plus identity verification, payment protection, hourly tracking, and project funds for security.

  • Define deliverables such as working pipeline components for retrieval, prediction, and final ranking stages.
  • Require submission of model training code and evaluation reports integrated into your existing data pipelines.
  • Establish performance improvement targets based on initial experimentation and ongoing production monitoring results.

Upwork is not affiliated with and does not sponsor or endorse any of the tools or services discussed in this article. These tools and services are provided only as potential options, and each reader and company should take the time needed to adequately analyze and determine the tools or services that would best fit their specific needs and situation.

The rates and information provided in this article are based on current data and industry sources available at the time of publication. Freelance rates can vary depending on factors such as experience, location, project scope, and market conditions. Readers are encouraged to conduct their own research to confirm current rates and trends, as this information may change over time.

How much does hiring a Recommender systems specialist cost?

$500-$1,500 per project is a typical range for focused Recommender systems specialist work. Final pricing depends on scope, technical complexity, required integrations, source-material quality, revision needs, and the freelancer's experience level.

Data preprocessing and feature engineering

$500-$1,200/project

Entry-level to mid-level
  • Processed user and item interaction data ready for modeling
  • Documented feature definitions and transformation logic
  • Summary of data quality checks and missing value handling

Offline model evaluation

$1,200-$2,500/project

Mid-level
  • Comparison of candidate generation policies using ranking metrics
  • Scripts to reproduce offline testing and metric calculation
  • Recommendation for the best performing baseline approach

Candidate generation pipeline

$2,500-$4,500/project

Mid-level to senior-level
  • Code to generate initial item candidates from large catalogs
  • Vector representations for users and items based on interactions
  • Unit tests verifying retrieval logic and data flow

Scoring and re-ranking system

$4,500-$7,000/project

Senior-level
  • Trained model to score and order candidates by relevance
  • Rules applied during re-ranking to meet business requirements
  • Service interface for real-time scoring requests

End-to-end production deployment

$7,000-$12,000/project

Expert-level
  • Live recommendation engine integrated with application infrastructure
  • Metrics tracking system performance and prediction drift
  • Documentation for maintenance, updates, and troubleshooting

Frequently asked questions

Is hiring a Recommender systems specialist worth it?

For most businesses, yes: hiring a Recommender systems specialist is worthwhile. These experts build custom ranking pipelines that surface relevant items to users based on their specific behavior and preferences. They validate model performance through offline testing before deployment to reduce the risk of poor user experiences.

How do I evaluate Recommender systems specialist candidates?

Look for candidates who explain how they balance candidate generation with precise re-ranking under real-world constraints. Ask them to describe a specific instance where they improved a ranking metric like NDCG or precision at k through model iteration.

What data do I need to start a recommendation project?

You must supply historical user interaction logs such as clicks, purchases, or view times along with item metadata. Clean and structured feature data allows the specialist to train accurate models for candidate generation and scoring.

How long does it take to build a recommender system?

Initial prototype pipelines often take four to eight weeks to develop and validate offline. Full production deployment with monitoring and continuous optimization requires additional time for integration with your existing infrastructure.