Senior Applied AI Engineer · Agentic systems, developer tooling and forward deployed engineering · London, UK
I build agentic systems and developer tooling, ship them to production, and deploy them alongside the people who use them. I have 8+ years in software: computer-vision research for ISRO's Chandrayaan rover, CTO of a 10-person AI/AR startup, and founding engineer at three AI companies. I hold an MSc in Artificial Intelligence (Distinction) from the University of Surrey.
At Lendable, I am a Senior AI Engineer, forward deployed across engineering teams to measure and improve how they use AI. I modelled incident.io, Jira, Datadog and GitHub data into governed, OKR-level Snowflake metrics. On top of that data, I built the AI Hub: a full-stack dashboard that unifies GitHub, Jira, Claude and Codex telemetry. I instrumented the organisation's agent skills and cut developer workflow times by up to 10%.
Before Lendable, I shipped agentic AI at the production edge. At Klyft, I cut LLM token costs 10x and made the core agent 3x faster (60s → 20s). At Move By Vision, I took a healthtech platform out of critical failure: 100% of AI crashes removed, latency down 55%, and a GDPR / EU AI Act audit done.
Alongside client work, I build in the open. Nexus-MCP is a local code-intelligence server for AI coding agents, published on PyPI. Protash turns a plain-English business intent into a deployed prototype in under 5 minutes.
AI & LLM Systems
LLM APIs
ML
Languages & Frameworks
Data, Cloud & Infra
| Project | Description | Tech |
|---|---|---|
| Nexus-MCP | Local code-intelligence MCP server for AI coding agents. Fuses vector search, BM25 and a code graph to cut tokens on code lookups by 70–90%. 15 tools, 25+ languages, <350MB RAM, no API keys. On PyPI and the Glama MCP marketplace | Python · ONNX Runtime · LanceDB · tree-sitter · rustworkx |
| Protash | Describe a business intent and get a deployed prototype back in under 5 minutes. A 6-stage multi-agent pipeline generates a React app, then QA and review agents self-heal it over up to 5 correction rounds. Live demo | Next.js · TypeScript · DeepSeek · GCP Cloud Run |
| CR8 (private) | 3-agent LangGraph pipeline that turns a university curriculum into study guides, decks, AI-narrated video and Bloom's-taxonomy quizzes. A two-layer eval framework scores every output. 1,299 tests | LangGraph · FastAPI · React · Kokoro TTS · Cloud Run |
| JobScout (private) | Daily pipeline that pulls ~30,000 jobs from 13 ATS platforms and checks each employer against the UK sponsor register. A free keyword tier removes ~80% of jobs before the LLM tier, so LLM cost stays under $0.01/day | Python · LangGraph · Embeddings |
| Health Intelligence Engine | 9-node LangGraph pipeline that grounds supplement advice in NIH DSLD and PubMed through a Neo4j knowledge graph. A pharmacovigilance critic checks each recommendation against the user's allergies and medications | LangGraph · Neo4j · RAG · DeepEval |
| Knowledge Graph Agent | NLP agent that turns raw search history into a Neo4j knowledge graph, then generates personalised video with Gemini and Veo 3.1. 11 semantic categories extracted for under $1 in API cost | Python · Neo4j · GPT-5-nano · Gemini · Veo 3.1 |
| RevGeni Agent | Async multi-agent pipeline that crawls the web, qualifies companies, enriches contacts and writes leads into a GraphQL CRM | Python · Multi-agent · GCP · GraphQL |
Lendable Senior AI Engineer · Jul 2026 – Present
- Forward deployed across engineering teams to measure and improve how they use AI
- Modelled incident.io, Jira, Datadog and GitHub data into governed, OKR-level Snowflake metrics for leadership
- Built the AI Hub: a full-stack dashboard that unifies GitHub, Jira, Claude and Codex telemetry into one view
- Cut developer workflow times by up to 10% through agent-skill instrumentation and automatic ticket-to-skill detection
- Built a company-level cycle-time model that attributes each PR to a team from historical contribution share, with a manager-facing correction loop
- Shipped on Terraform/Terragrunt, GitHub Actions and Argo CD, with Okta SCIM/OIDC authentication
Move By Vision Founding Engineer (Pro Bono) · Dec 2025 – May 2026
- Took over a Flutter + GCP healthtech platform in critical-failure alpha
- Built a guardrail and validation layer across the Vertex AI integrations and removed 100% of AI crash failures
- Cut generation latency by 55% (15s → 7s) and completed a GDPR / EU AI Act compliance audit
Klyft Technologies Founding AI Engineer · Sep 2025 – Nov 2025
- Built the full agentic stack for a coaching platform: multi-agent orchestration, RAG guardrails, memory, observability and CI/CD on GCP
- Cut LLM token costs 10x through prompt optimisation and caching
- Made the core workout-calibration agent 3x faster (60s → 20s), with 100% pilot uptime and 100% safe guardrailed outputs
Scan and Sew Inc. Founding Engineer · Oct 2024 – Aug 2025
- Shipped a Designer AI Assistant (RAG over design documents) and a real-time manufacturing tracking platform
- Evaluated custom Gaussian Splatting against third-party APIs and chose the API, which saved months of development
Earlier roles
AIMAGE Technologies · Chief Technology Officer · Jan 2020 – Aug 2023
- Founded the company and grew the team to 10 across engineering, product and design
- Built a custom vectoriser and a graph-based vector database for visual search, live on a retail client's storefront
- Built a virtual try-on engine (OpenCV, PyTorch, Unity3D) on AWS (Lambda, ECS, Neptune, DynamoDB, Kinesis)
Mage Ventures · Software Consultant · Oct 2019 – Mar 2021
- Implemented a cross-platform payment gateway for Walmart-owned PhonePe
- Built a computer-vision checkout engine for grocery retail
Cellstrat Inc. · AI Consultant (Founding Team) · Oct 2018 – Oct 2020
- Delivered AI/ML work for TVS Motors, Target, Airbus and Volvo in PyTorch, TensorFlow and Keras on AWS
- Delivered a workshop at ODSC on enterprise AI and deep learning to 100+ professionals
EndGate Global · Full Stack Engineer · Jul 2017 – Jul 2018
- Built bulk food-delivery order workflows in Java, Spring, Hibernate and MySQL
ISRO · AI Researcher Intern · Nov 2016 – Apr 2017
- Researched mono-vision depth detection for the Chandrayaan moon rover
- MSc, Artificial Intelligence (Distinction) — University of Surrey, United Kingdom (2024)
- Master's in Computer Applications (First Class) — Christ University, India (2017)
- Bachelor's in Computer Applications — Bangalore University, India (2013)
Certifications: Claude Code in Action, Model Context Protocol, Agent Skills, Building with the Claude API (Anthropic) · LangGraph & LangSmith (LangChain Academy) · Deep Learning & Computer Vision (Stanford)
Publication: Modern Machine Learning Approaches for Robotic Path Planning (peer-reviewed journal)