I turn complex workflows and promising AI prototypes into production-ready software platforms.
𝟭𝟵+ 𝗬𝗲𝗮𝗿𝘀 𝗼𝗳 𝗘𝘅𝗽𝗲𝗿𝗶𝗲𝗻𝗰𝗲 • 𝟯𝟬+ 𝗣𝗹𝗮𝘁𝗳𝗼𝗿𝗺 𝗠𝗼𝗱𝗲𝗿𝗻𝗶𝘇𝗮𝘁𝗶𝗼𝗻 𝗣𝗿𝗼𝗷𝗲𝗰𝘁𝘀 • 𝟯𝟬𝟬,𝟬𝟬𝟬+ 𝗣𝗿𝗼𝗱𝘂𝗰𝘁𝘀 𝗠𝗮𝗻𝗮𝗴𝗲𝗱 • 𝗟𝗲𝗱 𝗧𝗲𝗮𝗺𝘀 𝗼𝗳 𝘂𝗽 𝘁𝗼 𝟱𝟬 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗲𝗿𝘀
An AI agent or vision model that only works in a demo is not a product. I build agentic AI and computer vision systems that hold up in production—and the Spring Boot, Python, data, and cloud infrastructure they need to run at scale.
𝗪𝗛𝗔𝗧 𝗜 𝗖𝗔𝗡 𝗛𝗘𝗟𝗣 𝗬𝗢𝗨 𝗕𝗨𝗜𝗟𝗗
→ 𝗔𝗴𝗲𝗻𝘁𝗶𝗰 𝗔𝗜 & 𝗟𝗟𝗠 𝗦𝘆𝘀𝘁𝗲𝗺𝘀
Multi-agent architectures, tool-using agents, AI workflow automation, and production integrations using OpenAI, Claude, and Gemini.
→ 𝗥𝗔𝗚, 𝗙𝗶𝗻𝗲-𝗧𝘂𝗻𝗶𝗻𝗴 & 𝗘𝘃𝗮𝗹𝘂𝗮𝘁𝗶𝗼𝗻
Hybrid BM25 and vector retrieval, re-ranking, GraphRAG, Pinecone, Weaviate, and Qdrant, plus LoRA and QLoRA fine-tuning with Unsloth, Axolotl, Hugging Face TRL, and Crusoe—with hallucination control using RAGAS, DeepEval, and LangSmith.
→ 𝗖𝗼𝗺𝗽𝘂𝘁𝗲𝗿 𝗩𝗶𝘀𝗶𝗼𝗻 𝗮𝗻𝗱 𝗩𝗶𝗱𝗲𝗼 𝗣𝗿𝗼𝗰𝗲𝘀𝘀𝗶𝗻𝗴
Object Detection and Tracking, Pose Estimation, Anomaly Detection, CCTV Monitoring and Surveillance, Facial Recognition, Medical Imaging, 3D reconstruction, Diffusion models.
→ 𝗘𝗱𝗴𝗲 𝗔𝗜 & 𝗥𝗲𝗮𝗹-𝗧𝗶𝗺𝗲 𝗜𝗻𝗳𝗲𝗿𝗲𝗻𝗰𝗲
Model optimization with ONNX and TensorRT for NVIDIA Jetson Nano/Orin and Hailo, for real-time inference.
→ 𝗦𝗮𝗮𝗦 𝗣𝗹𝗮𝘁𝗳𝗼𝗿𝗺𝘀 & 𝗣𝗿𝗼𝗱𝘂𝗰𝘁 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴
Multi-tenant platforms and MVPs from data model to deployment using Java, Spring Boot, Python, FastAPI, Django, PostgreSQL, and Kafka on AWS, GCP, and Azure.
→ 𝗔𝗣𝗜 & 𝗦𝘆𝘀𝘁𝗲𝗺𝘀 𝗜𝗻𝘁𝗲𝗴𝗿𝗮𝘁𝗶𝗼𝗻
ERP, CRM, ecommerce, and healthcare integrations—including connecting AI agents securely to systems you already run.
→ 𝗦𝗼𝗹𝘂𝘁𝗶𝗼𝗻 𝗔𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲 & 𝗠𝗼𝗱𝗲𝗿𝗻𝗶𝘇𝗮𝘁𝗶𝗼𝗻
Architecture assessments, technical roadmaps, build-vs-buy decisions, due diligence, and modernization that cuts technical debt without a rewrite.
𝗦𝗘𝗟𝗘𝗖𝗧𝗘𝗗 𝗘𝗫𝗣𝗘𝗥𝗜𝗘𝗡𝗖𝗘 & 𝗥𝗘𝗦𝗨𝗟𝗧𝗦
✓ Architected an AI-powered pricing intelligence and catalog platform managing more than 300,000 industrial products, cutting quotation time from days to hours and improving search performance by 3×.
✓ Built an automated hunting platform to classify the shot as hit or miss, detect and track birds movement and classify their species.
✓ Built real-time detection and tracking for occluded, multi-camera environments—stable identity across viewpoints, behaviour classification from body motion—deployed across retail surveillance and aerial platforms.
✓ Built a real-time playing cards detection & tracking system using YOLO where each card was classified as player and banker. The model was optimized with ONNX and TensorRT for fast inference on Jetson Nano. Also, developed a MetaHuman based Virtual Receptionist to assist clients in bookings and guiding visitors and the solution was deployed on Jetson Orin.
✓ Contributed to enterprise healthcare interoperability: HL7 FHIR, Smile CDR, secure identity, and cloud health information exchange.
✓ Built an AI-enabled inventory and logistics platform for a bakery using Python, Django, Supabase, and OpenAI.
✓ Delivered enterprise workflow modules for a secure correctional management platform—approvals, case tracking, document management, and business rules.
𝗧𝗘𝗖𝗛𝗡𝗢𝗟𝗢𝗚𝗬 𝗦𝗧𝗔𝗖𝗞
𝗔𝗚𝗘𝗡𝗧𝗜𝗖 𝗔𝗡𝗗 𝗚𝗘𝗡𝗘𝗥𝗔𝗧𝗜𝗩𝗘 𝗔𝗜:
LangChain, LangGraph, LlamaIndex, GPT, Claude, Gemini, Qwen, Kimi, DeepseekR1, Hugging Face, RAGAS, DeepEval, LangSmith, Pinecone, Weaviate, Qdrant, LoRA, QLoRA, GraphRAG
𝗖𝗢𝗠𝗣𝗨𝗧𝗘𝗥 𝗩𝗜𝗦𝗜𝗢𝗡 & 𝗘𝗗𝗚𝗘:
YOLO, DeepSORT, BYTETrack, OpenCV, dlib, MediaPipe, MMLab, PyTorch, TensorFlow, Vision Language Models, Whisper, ONNX, TensorRT, NVIDIA Jetson (Orin, Nano), C++
𝗗𝗔𝗧𝗔, 𝗖𝗟𝗢𝗨𝗗 & 𝗠𝗟𝗢𝗣𝗦:
PostgreSQL, MongoDB, Docker, AWS, SageMaker, Bedrock, Lambda, GCP, Vertex AI, Azure, GitHub Actions, MLflow
𝗛𝗢𝗪 𝗜 𝗪𝗢𝗥𝗞
I start by understanding what needs to improve, who will use the system, what constraints already exist, and how we will know it worked. From there I stay hands-on through architecture, implementation, integration, deployment, and the debugging that comes after launch. Throughout, you can expect clear communication, documented architecture, and long-term ownership of what I deliver. I also provide fractional CTO-level support for architecture reviews, roadmaps, and vendor evaluations.
𝗡𝗘𝗫𝗧 𝗦𝗧𝗘𝗣
Send me two or three sentences about what you are building, its current state, and your timeline. I will reply with an honest read on scope, major risks, and the fastest safe path to produc
