Our goal: Turn every missed call into a booked job with smart voice agents, seamless SMS follow-ups, and rock-solid telephony. We're early-stage, fast-moving, and need someone who can own the voice backbone and make it bulletproof. The Challenge Our telephony setup (currently with Telnyx DIDs) has issues — inbound routing, call handling, and reliability problems that are blocking smooth missed-call recovery and voice agent automation. We need an experienced backend engineer to diagnose/fix it or migrate to Twilio (or a hybrid) so we can reliably: Capture missed calls and auto-respond with AI voice agents Enable instant booking via voice or SMS Support outbound reminders/confirmations Scale to hundreds of tradie users with high uptime What You'll Do Fix & Optimize Telnyx Integration: Troubleshoot DID inbound routing, SIP/WebRTC issues, call flows, and reliability. Ensure DIDs work flawlessly for inbound voice. Evaluate & Migrate (if needed): Compare and potentially switch to Twilio Programmable Voice for better ecosystem support, easier AI agent integration, or hybrid setup. Build Voice Automation Core: Inbound/outbound call handling (missed call → voicemail → AI agent → booking) Integrate with AI voice platforms (e.g., Retell AI, Vapi, or custom) Real-time call routing, IVR flows, transcription, and post-call actions (create jobs, send quotes) Backend Development: Node.js/Python (or your strong preference) services, webhooks, databases (likely Supabase/Postgres), API endpoints for the mobile app. Reliability & Compliance: Call recording consent, STT/TTS, error handling, monitoring, and scaling. Collaborate with the founder and mobile team to deliver end-to-end flows that feel magical for tradies. Requirements Strong experience with Twilio or Telnyx (both is a big plus) — Programmable Voice, SIP trunks, DIDs, webhooks, Media Streams. Proven track record building production voice/IVR/AI agent systems (missed call handling, conversational agents, booking flows). Solid backend skills (Node.js, Python, or Go preferred) + experience with real-time systems and webhooks. Familiarity with AI voice tools (Retell, Vapi, Bland, etc.) and telephony best practices. Ability to debug complex call flow issues quickly.

Job Title: Software Engineer (AI / Full-Stack) – SkinnyTax Location Remote Employment Type Part-time About SkinnyTax SkinnyTax is a tax-planning SaaS platform designed to help small business owners and high-income professionals make proactive, year-round tax decisions. We combine legal-grade tax strategy with modern software to simplify planning, improve compliance, and reduce surprises. We’re an early-stage, funded startup building real products for real users—and we move fast. Role Overview We’re looking for a Software Engineer to help build and scale the SkinnyTax platform, with a specific focus on developing an AI-powered chat feature that assists users with tax planning questions, intake, and guidance workflows. You’ll work closely with the founder to design, implement, and ship this feature end-to-end. This role is ideal for someone who enjoys ownership, problem-solving, and building practical AI features—not demos. What You’ll Do Design and build an AI chat system integrated into the SkinnyTax platform Implement conversational workflows for client intake, document guidance, and tax planning support Integrate LLM APIs and design prompt logic, guardrails, and structured outputs Build backend services and APIs to support AI chat functionality Ensure data security, privacy, and compliance for sensitive financial and tax information Improve performance, reliability, and scalability of AI-driven features Collaborate on product decisions, architecture, and roadmap priorities Write clean, well-documented, production-ready code What We’re Looking For 3+ years of professional software engineering experience Experience building AI-powered features (LLM APIs, chat interfaces, or similar) Strong backend skills (Node.js, Python, or similar) Experience with databases and API design Understanding of prompt engineering, system instructions, and response validation Ability to work independently in a fast-moving startup environment Nice to Have Experience building conversational AI or chat-based products Familiarity with fintech, tax, accounting, or compliance-heavy systems Experience implementing AI guardrails, logging, and monitoring Frontend experience (React or similar) Why SkinnyTax Build an AI feature used by real paying customers Direct collaboration with a founder who understands tax, law, and product needs Remote-first, flexible work environment Competitive compensation based on experience Opportunity to grow into a lead role as the product scales

We are seeking an experienced freelancer to set up an Odoo dashboard for our business. The ideal candidate will have a strong understanding of Odoo and its capabilities, as well as experience in creating customized dashboards. The project involves integrating various modules and ensuring seamless functionality. If you have a passion for Odoo and a knack for creating user-friendly interfaces, we would love to hear from you!

WHAT WE NEED We're building a meeting bot that joins Zoom sessions, deploys a single bot per breakout room, and pushes data to our backend. Recall.ai handles the Zoom infrastructure — your job is to build the integration layer between Recall.ai, Zoom, and our database. THE PROJECT IN ONE SENTENCE Build a working Recall.ai bot integration that joins Zoom meetings (including breakout rooms), captures meeting data, and delivers it to our backend, with clean documentation so the codebase is maintainable long-term. SCOPE OF WORK Recall.ai account setup and configuration Bot deployment logic: one bot per breakout room, dynamically assigned Zoom SDK integration where Recall.ai needs supplementing Webhook handling for bot lifecycle events (join, leave, error) Backend endpoint(s) to receive and store meeting data Basic error handling and logging Written documentation for long-term maintainability. One round of testing during a live session with our actual users WHAT'S OUT OF SCOPE (DO NOT QUOTE THESE) Front-end UI design Membership site or member portal Mobile app Custom AI/LLM analysis of meeting content Anything beyond the bot integration itself TIMELINE 4-6 weeks from kickoff. We have a detailed Phase 1 Scope of Work document (DOCX) ready to share with selected candidates. YOUR EXPERIENCE Required: Hands-on experience with Recall.ai (please mention specific projects in your proposal) Strong Zoom SDK / Zoom API background Backend integration experience (Python or Node.js) Webhook handling and event-driven architecture Clear written English communication Strongly preferred: US-based for time zone overlap during testing Experience with multi-room or breakout-room scenarios Examples of past meeting-bot work in your portfolio HOW TO APPLY In your proposal, please answer these three questions briefly: Have you worked with Recall.ai before? Link or describe one project. What's your typical approach to handling Zoom breakout room scenarios? What's your availability over the next 4-6 weeks? Generic "I can do this!" proposals will be skipped. We'll send the full Phase 1 Scope of Work document to candidates whose proposals demonstrate real Recall.ai familiarity. ABOUT US We run a live group practice program with 10-12 simultaneous breakout rooms per session. The bot integration is Phase 1 of a multi-phase build, and we're looking for someone we can work with long-term across future phases if this goes well.

About the Engagement Zeluca is a govtech AI company building ZAAP — a process intelligence and AI governance platform for federal agencies. We are looking to staff a government client engagement focused on Perplexity-powered enterprise automation, agentic workflow development, and AI-native tooling for a large US government organization. We need a hands-on Junior Prompt Engineer to join the delivery team and take ownership of prompt testing, iteration, and quality validation across a portfolio of LLM-powered automations. You will work within frameworks designed by a Senior Prompt Engineer and execute across a range of real automation workflows — not toy demos. This role is right for you if: • You have shipped working LLM automations — even small ones — and can show your work • You think carefully about output quality and know what "good" looks like for a given task • You're comfortable running structured tests, documenting failures, and iterating systematically • You can explain prompt design choices clearly in writing and in conversation • You are US-based and available to support a government client engagement (no clearance required at this stage) WHAT YOU WILL DO Prompt Development & Iteration • Implement and iterate on LLM prompt templates across Perplexity, GPT-4-class models, and open-weight endpoints • Run structured A/B comparisons, temperature sensitivity tests, format compliance checks, and output length validation • Maintain a versioned prompt library and apply change management protocols as templates evolve • Document test results, failure modes, and edge cases in shared registers Agentic Workflow Support • Configure and test LangChain / LangGraph tool-use nodes under Senior Engineer direction • Validate agentic pipeline outputs against acceptance criteria and HITL escalation thresholds • Assist in integration testing of Perplexity search results inside multi-step reasoning chains • Log and triage output quality issues; escalate unresolvable failures with clear documentation Quality & Documentation • Score LLM outputs against rubrics covering accuracy, relevance, tone, and format compliance • Produce prompt run logs, quality summary reports, and regression results each sprint • Update prompt SOPs and inline documentation as templates evolve • Flag PII / sensitive data exposure risks during output review Stakeholder & Delivery Support • Support demonstration preparation for government end users and program managers • Assist in writing training materials for non-technical staff using LLM-powered tools • Respond to day-to-day prompt refinement requests from government users under Senior Engineer oversight WHAT WE ARE LOOKING FOR Must-Have ✓ 1-3 years of hands-on work with GPT-4-class LLMs, Perplexity, or equivalent — not just API familiarity ✓ Demonstrable prompt engineering experience: zero-shot, few-shot, chain-of-thought, structured output (JSON/Pydantic), and instruction following ✓ Ability to diagnose common LLM failure modes: hallucination, format drift, verbose drift, instruction following failures ✓ Python scripting to automate prompt test runs, parse outputs, and log results ✓ Clear, structured written communication — your test reports and escalation notes need to be unambiguous ✓ US-based (required for this engagement) Strong Advantage • Experience with Perplexity API — search-augmented prompting, citation handling, query structuring • Exposure to LangChain, LangGraph, or any agentic framework — even a personal project counts • Background in QA, structured test engineering, or software testing methodologies • Familiarity with government / defense / federal agency environments or data handling requirements • Experience with Langfuse, W&B, or any LLM observability / tracing tool What We Are NOT Looking For: • Theorists — if your experience is primarily coursework or reading about prompting, this is not the right fit yet • Generalist AI enthusiasts without structured prompt testing experience • Candidates without the ability to work in a structured, documented delivery environment • Candidates requiring significant ramp-up time — this engagement starts active work immediately ENGAGEMENT DETAILS Type: Contract — Time & Materials Rate: $90 - $110 / hour (USD) Hours: 80 to 160 hrs/month; part-time options available for strong candidates Location: Remote — US-based required, mid-west preferred Start: Within 7-14 days of selection Duration: Initial 3-month engagement; extension based on performance and client scope Clearance: Not required at this stage; clean background check required before start Reporting: Senior Prompt Engineer (Zeluca) + Engagement Leader Sprint Cadence: Weekly check-in; bi-weekly delivery review with client team HOW TO APPLY — PROPOSAL QUESTIONS To help us evaluate fit, please address the following in your cover letter. These are not tasks to complete — we want to understand your thinking and experience based on work you have already done. Question 1: Prompt Design & Testing Describe a prompt you designed for a real LLM automation. What failure modes did you encounter, how did you diagnose them, and what did you change to fix them? Be specific about the model, the task, and the output quality criteria you were working against. Question 2: Output Quality Scoring How do you evaluate whether an LLM output is "good enough" for production? Describe the rubric or criteria you use. Have you ever scored outputs systematically across multiple prompt variants? What did you learn? Question 3: Agentic or Search-Augmented Experience Describe any hands-on experience with LangChain, LangGraph, Perplexity API, or similar tools — even personal or side projects. What did you build, what worked, and what would you do differently? Please also include: a brief cover note (3-5 sentences max) on your relevant experience and the most complex LLM automation you have shipped, links to relevant prior work (GitHub, portfolio, past LLM projects), and confirm you are US-based and available within 7-14 days. Questions? Message via Upwork before submitting. We respond to all messages within 1 business day. Zeluca | zeluca.com

We're a productized AI service running an agent runtime — one isolated instance per customer — on a stack of FastAPI, Next.js 14, Postgres, and Docker. We're past MVP and into production hardening: real customers, real revenue, real SLAs. Inherit and improve the existing FastAPI control plane (routes, async SQLAlchemy 2.0 + asyncpg, Pydantic, JWT, Stripe billing, APScheduler) Tighten the Next.js 14 / React 18 / Tailwind / Clerk auth frontend Improve security posture: per-tenant row-level security in Postgres, input validation, rate limiting Performance pass: SSE streaming, DB query optimization, caching layer Work alongside an AI agent engineer and a Twilio specialist (separate hires) Required: 7+ years full-stack production experience Deep FastAPI + async Python (NOT a Flask/Django person retrofitting) Next.js 14 App Router + TypeScript at production scale Postgres at scale — async drivers, RLS, query planning Docker, comfortable with Render or similar managed infra Comfortable inheriting a working codebase — you read before you rewrite Nice to have: Multi-tenant SaaS architecture experience Stripe metered billing Clerk or similar managed auth Sentry, observability Have inherited and improved an AI/LLM-adjacent platform before How we work: This is an ongoing engagement, not a one-off. 20–30 hrs/week. Async-friendly but US time zones strongly preferred for sync calls. We pay $80–150/hr depending on seniority. We are not the lowest bidder's client. Paid 4–8 hour trial task (real codebase work) before any retainer.

1. Project Overview This project builds a computer vision system that tracks whether a handheld object is inside or outside a user defined region on a person's head in real time. The system observes the user through a webcam, allows the user to define a region on their head once at the start, tracks that region as the user moves, and outputs a binary signal whenever a tracked object is inside or outside the region. The signal is transmitted to an Arduino over USB serial and can be used to control any device, such as turning a flashlight on when the object is in the correct region and off when it leaves. The previous developer completed Milestone 1, which covers webcam capture, frame rate monitoring, multi camera support, and a basic overlay system. The remaining work is organized into five milestones over an estimated nine weeks. 2. System Architecture The system runs as a single Python application on the user's computer. It captures frames from the webcam, runs all computer vision processing, and sends commands to the Arduino over USB serial. The Arduino runs a minimal firmware that reads those commands and switches an output pin accordingly. That output pin can drive any device the client chooses. The processing pipeline has six stages that run on every captured frame. First, the frame is read from the camera and corrected for lens distortion using stored calibration parameters. Second, MediaPipe Face Mesh detects 468 facial landmarks on the user's head. Third, the previously stored region anchor points are reprojected into the current frame using the new landmark positions, which is how the region follows the head as it moves. Fourth, a color based tracker locates the handheld object in the frame. Fifth, the system computes whether the object is inside or outside the region using a point in polygon test. Sixth, the resulting on or off signal is sent over serial. A separate safety supervisor runs alongside the main pipeline. It forces the output signal to off if the face cannot be detected, if the object cannot be tracked, or if the serial connection is lost. This guarantees the output is only on when the system is confident about the object's position This WHAT Im looking to be Completed a simple part of the project: Milestone 1: Camera Setup & Calibration Goal: Set up the camera and calibrate the system for accurate real world distance measurements. Steps: Set up and test the camera for real time video capture. Correct camera distortion using OpenCV and a checkerboard pattern. Calibrate the system using a reference object of known size and adjust dynamically based on the user's inter pupillary distance. Success Criteria: Achieve real world measurements with an accuracy within 5%. Correct camera distortion and dynamically adjust scale during use. I do have existing code that does cover this part:with basic webcam capture, FPS display, multi camera preview and overlay logic It is however missing this:calibration, head anchored region tracking, smoothing, object detection, Kalman filtering, Arduino communication and safety logic are missing.

We are looking for an experienced AI automation developer to build a private executive assistant named Jarvis for a business owner named Vince. Jarvis must operate as a professional, respectful, fast-moving executive assistant. The assistant will communicate with Vince through iMessage, access his Google Calendar, remember important information, send meeting reminders, and maintain local files/data on an office iMac. This is not a basic chatbot. We need a working AI assistant that can hold real conversations, remember context, anticipate needs, and protect Vince’s time. Core Requirements The assistant must: Communicate with Vince through iMessage on macOS. Store all data, memory, and files locally on the office iMac. Access Vince’s personal Google Calendar. Send Vince a message 20 minutes before meetings. Remember meeting times, preferences, important facts, and prior conversations. Use context from previous messages and stored memory. Start conversations professionally with: “Hello Sir. What do you need today sir.” Maintain a direct, respectful, professional tone. Avoid fluff, long explanations, repetition, and unnecessary questions. Understand that Vince has zero tolerance for wasted time. Validate Vince’s instructions and respond with useful answers quickly. Ask onboarding questions at first launch to learn Vince’s occupation, goals, priorities, communication preferences, daily routines, and assistant expectations. Be built in a way that can expand later into email, task management, document handling, and proactive reminders. Important Personality / Communication Rules Jarvis must be designed around Vince’s communication style: Direct. No fluff. No jargon. Lead with the answer. Never ask for information Vince has already provided. Protect his time, brand, relationships, and workflow. Jarvis should function as an executive personal assistant whose purpose is to remember everything so Vince does not have to repeat himself. Technical Scope The developer should be comfortable with: macOS automation. iMessage / Messages.app integration. Google Calendar API. Local file storage and local memory architecture. AI agent frameworks. Cron jobs or scheduler-based reminders. Secure credential handling. Local database or file-based memory. Python, Node.js, or similar automation stack. Optional: BlueBubbles, AppleScript, Shortcuts, SQLite, vector database, local LLM tools, OpenAI API, Claude API, or similar. There is already a macOS/iMessage path available using CLI-based message tooling, but we are open to the developer recommending the best reliable implementation. Existing iMessage automation concepts include sending, reading, and watching message history through macOS Messages.app tooling. Deliverables We need the developer to provide: Working Jarvis assistant installed on the office iMac. iMessage communication with Vince. Google Calendar integration. Automatic 20-minute meeting reminders by text. Local memory system. Local file/data storage structure. First-run onboarding question flow. Prompt/personality system for Jarvis. Basic admin documentation showing how to restart, update, and maintain the assistant. Security notes for credentials, permissions, and local storage. Testing checklist proving iMessage, memory, reminders, and calendar sync work. First-Run Intro Flow Jarvis should text Vince an introductory message and ask important setup questions such as: What is your primary occupation? What are your top business priorities right now? What meetings or events should I always remind you about? Who are your key contacts? What should I never interrupt you for? What should I always notify you about? What tone do you prefer from me? What daily reminders would make your life easier? What are your current goals for the next 30, 60, and 90 days? Ideal Candidate The ideal freelancer has built AI agents, personal assistants, calendar bots, local automation tools, or macOS/iMessage workflows before. We want someone practical who can build a reliable working system, not just create a demo. Please include: Similar AI assistant or automation projects you have built. Your recommended tech stack. How you would connect iMessage. How you would handle local memory. How you would secure calendar credentials. Estimated timeline. What you need from us to start.

About us Nucci HQ is an established coaching and consulting company. We run one-on-one and group coaching programs, a consulting arm, and a real estate division, with an active student community and a fast-growing content presence on YouTube and social. We move quickly, and we are investing heavily in AI to make every part of the business faster — and we are looking for the person who will build and own that. The role We are hiring a contract AI Agent & Automation Engineer to design, build, fix, and maintain the AI agents and automations that run across our business — sales, operations, finance, coaching, and content. You will work directly with our founder, Anthony, ship quickly, and see the impact of your work right away. There is a clear roadmap of work ready from day one: stabilizing automations we have already built, standing up an internal command-center dashboard, and shipping a library of single-purpose agents. This is a contract role billed hourly — a flexible, low-friction way for both of us to start, with room to grow as the work does. What you’ll do • Fix and stabilize existing automations so they run reliably. • Build an internal CEO dashboard that pulls sales, consulting, finance, real estate, and social metrics into a single source of truth. • Build workflow automations — meeting note-taking from call transcripts, a speed-to-lead responder, coaching-call tracking, and weekly accountability recaps. • Build a library of single-purpose AI agents across sales, operations, finance, and content — each with one clear job. • Connect agents to the tools we already run on — Slack, Google Workspace, Whop, Heartbeat, ManyChat, Granola, and Fathom. • Prototype new ideas quickly, then harden the ones that work into reliable, monitored systems. • Document what you build so the team can use it and trust it. What we’re looking for • Two or more years building software, with hands-on experience building LLM-powered apps or agents. • Strong Python and/or JavaScript / TypeScript. • Practical experience with LLM APIs (Anthropic, OpenAI, or similar) — prompt design, tool / function calling, and retrieval (RAG). • Strong, hands-on experience with n8n, webhooks, and APIs — building, connecting, and debugging integrations. This is essential. • Comfortable reviving or debugging automations that someone else started. • Good judgment about when an agent is reliable enough to ship — and how to test for it. • A clear communicator who is self-directed and comfortable owning projects end to end. Nice to have • Experience with agent frameworks and orchestration (LangChain, LlamaIndex, MCP, CrewAI, or similar). • Familiarity with our stack — Slack, Whop, Heartbeat, ManyChat, Granola, or Fathom. • Front-end skills for building dashboards and internal tools. • Experience deploying and monitoring production AI systems. • A portfolio, GitHub, or side projects we can look at. Details Location: Fully remote. Type: Contract, billed hourly. Rate: $70–$110 per hour, depending on experience. Hours: Flexible — we will agree on a weekly range based on your availability. Start date: As soon as possible. How to apply Send a short note about a recent AI agent or automation you built and what it does, your resume or LinkedIn, and links to any code, demos, or projects. No long cover letter needed — we care most about what you have built.

I am looking to partner with a competent dev team (individual or team) to launch AI implementation solutions for small and medium-sized businesses. As a sales professional, I aim to provide innovative solutions that enhance business operations. The ideal candidate will have experience in AI technologies and a strong understanding of business needs.