Engineering Leader · Technical Co-Founder
Things I've built.
Production voice AI agents, LLM pipelines, MCP servers, and event-driven backends serving 40,000+ locations, plus mobile apps live on the App Store and Play Store. Active Claude Code production user, heavy n8n automator, and long-time tinkerer with anything that puts an LLM on the critical path.
00 · Latest
1st place · 18 August 2026
Won Adaption Labs's AutoScientist Challenge in Data & Visualization
PolyChart scores how much a chart lies, giving every distortion an exact lie factor rather than a yes/no label. Built on a controlled study of more than 200 fine-tuning jobs, driven through a headless client I wrote before the platform's official API shipped. Entered in all seven categories; placed first in this one.
01 · Production
Production Systems
Commercial products shipped end-to-end. Voice AI, AI career prep, education ops, real-time golf — all live or in production.
AI Career Prep Platform
Senior Software Engineer, Forward Deployed & Team Lead · VoiceQube
- Real-time AI voice mock interviews — Pipecat + LiveKit + Claude + ElevenLabs + Deepgram
- LLM-driven resume tailoring per JD with RAG over candidate documents
- Job matching, application tracker, interview prep, upskilling paths
- Resume parsing pipeline: Tesseract OCR + pdf-parse + mammoth
- Multi-payment integration: Razorpay, Stripe, Apple Pay
AI Voice Caddie ft. Andy North (2× US Open Champion)
Senior Software Engineer, Forward Deployed & Team Lead · VoiceQube
- Real-time voice caddie via LiveKit + WebRTC stack
- MongoDB geospatial indexing for live GPS course position
- AWS Lambda + MediaConvert for shot video capture and processing
- Production-grade across 40,000+ golf courses (event-driven backend)
- Voice + intent stack with Anthropic Claude in the loop
Academic Operations Platform
Senior Software Engineer, Forward Deployed & Team Lead · VoiceQube
- 447 of the repo's 2,688 merged PRs are mine; most commits of any contributor
- Admissions CRM with an AI voice agent that calls new enquiries in five seconds (~220k LoC)
- IPE: AI grading of handwritten answer sheets, 12,847 marks reviewed by faculty in prod
- Hono API (152 route modules), Drizzle over Postgres — 323 tables, 216 migrations
- 125,299 questions in the bank, 78% AI-drafted, none published without faculty approval
AI Admissions CRM · a call within 5 seconds of every lead
Senior Software Engineer, Forward Deployed & Team Lead · VoiceQube
- AI voice agent calls every new lead within five seconds and fills the enquiry form from the conversation
- Click-to-call from the browser, and bridged calls where the agent dials the RM, then the parent, and listens to both
- Live coaching mode: suggestions from the live transcript plus sentiment, on the RM's screen
- Brand Factory: a Lovable-style editor over each institution's assets that generates landing pages, links and AI ad creatives
- Meta and Google ad accounts connected over OAuth; Twilio and Plivo SIP trunks with numbers per institution
Assist-only AI for boutique hotels · PMS integrations
Full-Stack & AI Engineer (contract) · HotelClaw.ai
- One PMS sync runner, five adapters: an Oracle OPERA export translator live at the pilot resort, plus Opera Cloud (OHIP), Mews, Cloudbeds and Apaleo
- LINE and WhatsApp webhooks, HMAC-verified, written to an event log
- Most of the LLM intent router: it drafts and routes, and sends high-risk guest messages to a person
- Tenant isolation as a CI check: row-level security on every table, exceptions taken from 54 to zero
- Engineering baseline from day one: Biome, PR gates, CI, secret scanning
- Architected end-to-end across iOS, Android, web, vendor ecosystem
- ML-powered planning workflows and intelligent vendor matching
- Viral QR photo-sharing for live event capture
- Live on App Store and Google Play; multi-market customer base
- Established engineering culture, CI/CD automation, technical roadmap
- Personalized AI chat coach with server-side key handling (Apple-compliant)
- Care-pathway with goal setting, progress tracking, motivation, flexibility
- ICMR STS-funded research project; psychometric analysis and UX study
- Award-winning: AIDEC 1st place, 3rd place, plus ICMR Short-Term Studentship
- Open-source react-native boilerplate ancestry (rodriguescarson-maeda)
- Care-pathway design adapted from SmokeMukti to oral-health × HIV intersection
- Live on both App Store and Google Play
- Built on Expo / React Native with Redux Toolkit and React Navigation 6
- iOS build via EAS; Apple-compliant data and login flows
Career Counselling Platform
Senior Software Engineer, Forward Deployed & Team Lead · VoiceQube
- 13+ feature modules: llm, payment, resume, counselor, code-compiler, forum, jobs, feedback, user, admin, auth
- Full Prometheus + Grafana + Sentry observability stack
- Cross-product integrations with Phiny.ai
- LLM routing through OpenRouter (Claude / GPT-4 / Gemini)
- Generative dashboard — agent picks chart types, KPIs, filters from a natural-language brief; widgets stream onto the canvas as tool calls return
- CopilotKit Studio — meta-tool where the agent authors useCopilotAction definitions live and previews them in a sandbox
- Voice onboarding — hands-free fitness profile via Web Speech API; transcript streams to CopilotKit which calls per-field tools
- React Native + Expo client over the wire-only AG-UI protocol via @assistant-ui/react-native — same agent backend, native surface
- Persistence layer, per-exercise progress tracker, curated GIPHY map, editorial dark design system, global studio nav
- AI-native architecture — Groq for fast couples-specific prompts, Claude for safety-critical paths; AI features opt-in and configurable
- Supabase Realtime for live shared albums, notes, and couple-specific collaboration
- Multilingual from day one — 8 locales (en, es, pt, hi, tl, ar, ko, fr) including RTL
- Rich feature set: albums, love letters, games, mood tracking, date planning, gift ideas, mediation, memory books, weekly story generation
- Web on Cloudflare (Astro + Workers) + mobile via Expo over a shared Postgres backend with edge functions
- AI mental-health companion on Groq (Llama 3.3 70B) with a guardrailed system prompt that refuses diagnoses, medication advice, and harmful content
- Crisis-signal detection auto-surfaces SOS resources; prompt-injection hard refusal and per-message clamping at the edge
- Breathing and 5-4-3-2-1 grounding exercises animated with Reanimated 4
- Daily mood check-ins with streak tracking and trend visualization
- Privacy-first — chat, mood, journal, and check-in data stored locally, with one-tap erase
- Google Maps integration — geolocation, clustered nearby discovery, and direct navigation
- AI dining concierge on Groq over Vercel Functions, with seeded places injected as RAG context
- Influencer-curated spots with a Supabase seed pipeline; photo uploads, reviews, and saved lists
- Google OAuth + Apple Sign-In across Web, iOS, and Android
- RevenueCat Pro tiers for consumers + Stripe vendor/brand-deal checkout
- AI Snap — photo → macros via a Supabase Edge Function calling Claude (analyze-food)
- Barcode scanning over Open Food Facts for instant nutrition lookup
- Apple Health (HealthKit) + Android Health Connect two-way sync
- Onboarding-driven calorie budget with macro rings and gifted-charts trends
- Anonymous bootstrap → email / Apple auth, RevenueCat subscriptions, i18n from day one
02 · Building with Claude · AI · Automation
The agentic dev workflow
Claude Code in the daily build/test/ship loop, MCP servers in production, n8n orchestration, and real-time AI avatars. AI is the work, not a side project.
n8n LLM Orchestration · Ôdasie
Heavy n8n automation for LLM agent orchestration, API integrations, and approval workflows. Cut release cycles from weeks to hours via AWS infrastructure automation (EC2, S3, SQS, Lambda, Bedrock) plus Netlify, Coolify, GitHub Actions.
MCP Servers in Production
6+ months of production MCP usage on real workloads at Ôdasie, then formalized into a paper on MCP Server Architecture Patterns (arXiv preprint).
Real-Time AI Avatars
Built bi-directional voice + video + data avatar experiences using LiveKit + WebRTC at Ôdasie. Live talking AI avatars wired into customer onboarding and demos.
03 · Open Source
Open Source & Tools
Reference implementations and tools — the voice AI server you can clone, a competitive-programming workspace, real-time document chat.
Qubekit Voice AI Server
Open-source LiveKit + Pipecat + Claude + ElevenLabs + Deepgram FastAPI reference implementation — SSE streams, JWT, runtime data exchange, React test client. The voice stack you can clone and ship.
RealStream ChatDoc
Real-time document intelligence platform — Kafka-based ingestion, AWS Glue + Athena ETL, vector-based document chat over Pinecone + Algolia + OpenAI.
04 · Recognition
Recognition
Peer review service and academic awards.
1st place · AutoScientist Challenge, Data & Visualization
Adaption Labs · $4,000 first prize
Won the Data & Visualization category of Adaption Labs' AutoScientist Challenge (Part 2) with PolyChart — a model that scores how much a chart lies, trained against an exact lie factor rather than a yes/no label. Entered all seven categories off one model and a 176-job controlled study. Full write-up at /adaption-autoscientist.
Certificate of Reviewing
European Journal of Computer Sciences and Informatics
Evaluated a manuscript as part of the peer review process — contributing to research quality in computer science and informatics.
Best Outgoing Student Award
Don Bosco College of Engineering, Goa
Chairman of C-Codes (Students' Council) — organised INSPIRUS 2K21, two-day national tech event with 300+ participants.
Runners-up · Rio Designeiro UI/UX Designathon
Technix 2023 · Goa College of Engineering (Govt. of Goa)
Placed 2nd designing a furniture-selling app end-to-end with my team (Siddhesh Kamath, Germano D'Costa) — a unique, user-first solution delivered under competition time pressure.
05 · Earlier
Earlier Work
Enterprise IoT before voice AI; non-profit GenAI education before founding.
Siemens Enterprise IoT
Enterprise IoT desktop applications — 20k+ LoC, 1,000+ Jest tests, MQTT pub/sub, concurrent schedulers achieving ~3× performance improvement.
More on the blog — AI, voice systems, MCP, dev tools.
Available · London / UK · senior AI, AI FDE, contract
Building something with AI?
Voice agents, MCP servers, LLM pipelines, agentic workflows — pick a slot, drop a message, or send your email and I'll reply within a day.
Replies within ~24 hours · London, UK · on-site, hybrid or remote