résumé · 2026.11
Matheesan Manokaran
Senior Software Engineer · Marketplace Platforms · AI Systems
summary
Senior software engineer with seven years of AI/ML delivery. Currently building WayTrade’s AI-powered marketplace for biofuels, feedstocks and critical materials on a small product team. Before that, five years building production LLM agents, model routing, knowledge bases and RAG, evals and the agentic tooling around them. Strongest at the gap between green tests and production behaviour — and at choosing the tool that shortens the loop without hiding the failure.
experience
WayTrade
Senior Software Engineer · Marketplace Platform
Nov 2026 — Present
Remote
- Building the AI-powered marketplace for biofuels, feedstocks and critical materials on a small engineering team.
- Connecting commercial, operational and compliance workflows on one trade record — contracts, logistics, inventory, documents and finance.
- Applying AI where it removes friction: matching supply and demand, surfacing exceptions and recommending the next action.
Replicant
Senior Software Engineer · Delivery / Forward-Deployed AI
Oct 2021 — Oct 2026
Remote · Toronto
- Built the org-wide agentic workbench solo: one engine connecting multiple data silos and workstreams through plugins and MCPs, with DeepSeek/Claude overnight planning, scoping, building and JIRA organization behind gates.
- Took enterprise LLM voice agents from an empty repo to production: state machines, tool calls, auth, fallbacks and hypercare.
- Built the platform layer: provider routing and fallbacks, knowledge bases and RAG, eval harnesses, guardrails and observability.
- Caught launch-blocking auth, wrong-record and first-time-user defects before they reached users; turned them into acceptance tests and release gates.
- Led a stalled enterprise migration to 100% cutover in five weeks; rebuilt routing, hand-off payloads and a CI test gate.
- Shipped AI coding skills and agentic tooling; drove GitHub Copilot, Cursor and Claude Code adoption across engineering.
- Worked across frontier and open-weight providers; treated model and prompt upgrades as releases with regression evals.
CGI
Software Engineer — Emerging Tech
2019 — 2021
Toronto
- Built OCR + classical NLP document pipelines before LLMs made extraction a prompt.
- Took models from notebooks to reproducible deployments with containerisation, CI/CD, monitoring and rollback.
- Ran applied-AI discovery: rapid prototypes, build-vs-buy analysis and handoff to operations.
Avaya
Software Engineering Intern — Professional Engineering Year
2018
Toronto
- Built test harnesses and log tooling for SIP/VoIP call flows.
- Traced latency and state across real-time systems that could not be restarted casually.
selected work
The agentic workbench
One engine for the whole engineering org: plugins and MCPs connecting multiple data silos and streams of work, with DeepSeek/Claude overnight planning, scoping, building and JIRA organization behind gates. Built solo with my own harnesses.
The production voice agent
Empty repo to live, multi-turn agent: authentication, eligibility checks, atomic transactions, fallbacks, and three launch blockers caught before users met them.
Silent green
Collected 16 real AI-authoring failures; 6 were “silent green” — change made, tests passed, defect shipped. Turned them into eval scenarios with must / must-not assertions and five lint rules accepted for the platform linter.
The migration rescue
Rebuilt transfer routing from a production snapshot, expanded hand-off payloads, and made a product-facing test suite a CI merge gate. Cut over to 100% of traffic in five weeks.
The prototype that lost
Benchmarked a Cloud Run replacement for a high-volume integration platform: p95 cold start 401 ms against a 300 ms target, ~99% lower modelled cost — and recommended the competing platform on the evidence.
skills
- LLM & agents
- Tool-calling agents · state machines · prompt/state design · model upgrades · ASR/TTS pipelines · evals · MCP
- AI platform
- Model routing · fallbacks · knowledge bases · RAG · vector search · embeddings · citations
- marketplace
- Trade workflows · offers · logistics · inventory · documents · compliance · finance
- AI tooling
- GitHub Copilot · Cursor · Claude Code · MCP · LangChain · Ollama · Whisper · Open WebUI
- languages
- TypeScript · Python · SQL · Bash
- cloud & infra
- GCP (Cloud Run, BigQuery) · Terraform · Docker · CI/CD · observability
- integrations
- Genesys · NICE · Amazon Connect · Salesforce · REST / OAuth2
- practices
- AI-native SDLC · forward-deployed delivery · incident response · postmortems · technical writing · mentoring
education & recognition
BEng, Computer Engineering
Ryerson University (now Toronto Metropolitan University) · 2015 — 2019
Final-year ML project: Twitter data → interaction graphs → community structure.
Earlier foundations
Document intelligence (CGI, 2019) · MLOps runway (CGI, 2020)
Real-time comms (Avaya, 2018) · graph communities (TMU, 2019)