Resume

AI systems engineer · Founding AI at an early-stage travel-tech startup · 12+ years building production systems

Experience

Early-Stage Travel-Tech Startup • Bangalore, India • 2026 - Present

Built the company's flagship experiential-navigation product from scratch — guided, narrated city journeys (H3 geospatial indexing, real-time routing, waypoint reasoning) for metros like New York and London
Architected a hybrid multi-agent engine generating grounded, place-aware stories and navigation cues at scale (~19K waypoints), served to the live mobile app
Built a visual-intelligence system pairing an open-source VLM (UI-TARS) with an LLM to refine POI physical coordinates from street imagery — sharpening last-mile accuracy of story delivery
Designed the POI knowledge-graph / RAG foundation — agentic multi-source discovery, entity resolution, and semantic dedup — as the ground-truth layer the product is built on
Built an agentic boundary-identification layer fusing open boundary datasets (OSM/Overture) with visual intelligence to resolve true place boundaries — 89% clean-polygon accuracy, zero misses — productionized as a durable package
Designed an OpenTelemetry observability standard for GenAI (traces, latency, real per-provider cost across ~140 call sites on AWS); built an event-driven telemetry engine in Go

Diligenta • Edinburgh, UK • 2024 - 2026

Architected and shipped an AI-agentic research engine chaining Llama-3 models with vector-grounded search, pushing factual-accuracy from 88% → 97%
Designed a canonical entity model and H3-based conflation pipeline that merged multi-source POI data; lifted merge precision to > 95 %.
Centralized authentication by extracting IAM logic into modern API Gateway layer with standards-based token validation
Engineered a hybrid retrieval system by fusing FAISS(semantic) and BM25 (lexical) search with Reciprocal Rank Fusion (RRF). This approach improved grounding fidelity and reducing off-topic results.
Led cross-functional team establishing engineering standards, mentoring, automated testing, and observability

Diligenta • Edinburgh, UK • 2022 - 2024

Incorporated SQS a messages queuing layer to decouple microservices communication, enhancing system resilience and scalability under peak loads, improved average 90th percentile latency by 30%
Implemented adaptive MFA (RFC-6238 compliant) reducing authentication failures and improving device onboarding
Designed and shipped eSignature workflow with Docusign, reducing advisor turnaround by 50%
Shipped Apple Pay & Google Pay integrations, increasing digital payment adoption by 10%

Tata Consultancy Services • Bangalore, India • 2018 - 2022

Built full SSO solution using Forgerock IDP with OAuth/OIDC flows, consent handling, and policy-based access
Migrated core backend infrastructure from AWS ECS → Azure AKS, improving cost efficiency
Implemented semantic document search using embeddings and Elasticsearch for legal documents

Tata Consultancy Services • Chennai, India • 2016 - 2018

Built high-throughput WebSocket messaging server handling 40M+ messages/day with low-latency fan-out
Designed AI-driven mentor–mentee recommender, improving match success by 30%
Developed MS Teams–integrated nudging platform using Bot Framework and adaptive cards

Advisory Board Company • Chennai, India • 2014 - 2016

Built HMAC-based service authentication libraries used across internal microservices
Developed extensible webhook framework enabling cross-system communication & CI/CD triggers

Amrita School of Engineering • India • 2010 - 2014