Bhavya Bansal
Applied AI Engineer & Engineering Leader
Full-stack engineering leader with 7 years building and scaling products from 0→1. Currently Sr. Software Development Manager at Traversia Technologies, shipping agentic AI systems, Model Context Protocol (MCP) integrations, GraphRAG, and high-concurrency backends handling 1M+ daily requests. Open to full-time and consulting roles — New Delhi / Remote.
- [email protected]
- New Delhi, India · Remote
- GitHub
- ibhavyabansal.com
Experience
Sr. Software Development Manager
- Architected Crux — high-concurrency API aggregator for 50+ hotel, DMC, and payment providers handling 1M+ daily requests at 99.9% uptime.
- Deployed hybrid search & RAG pipelines with sub-200ms retrieval latency for enterprise partners.
- Raised CI/CD and automated testing standards; reduced critical production bugs by 25%.
- Partnered with B2B clients to translate operational pain points into scalable architectures; onboarding +40%.
Head of Engineering
- Led 0→1 engineering for a B2B/B2C holiday platform scaled to 100K+ concurrent users.
- Cut client infrastructure costs 50% via AWS/Azure optimization and pipeline refactoring.
- Defined Hotels and Holidays product roadmaps with stakeholders via iterative prototyping.
- Mentored teams; sprint delivery velocity +30%.
Lead Developer (Frontend)
- Led MEAN stack frontend delivery; page load times −30% through prototyping with B2B clients.
Lead Software Developer
- Built GIS pipelines ingesting 2,000+ high-res images per batch into 3D terrain models.
- Distributed batch processing for 500+ concurrent mapping requests at 99.9% uptime.
Software Development Engineer
- Built Geo Mapper — real-time orthophoto visualization with live video and Google Maps API.
Founder
- Owned 0→1 product lifecycle from market validation through full-stack execution and beta users.
Selected projects
Travel Professor: Agentic Booking Concierge
Autonomous travel agent on Crux with multi-step reasoning for hotel discovery and itineraries. Custom MCP server bridging live B2B APIs to LLMs.
Adaptive Exam Engine & AI Grader
Mock-ETS adaptive testing with GraphRAG syllabus maps and a DSPy-compiled AWA grader aligned to 6-point rubrics.
LLM Evaluation & Factuality Pipeline
CI/CD-integrated evals for document agents — Context Precision, Faithfulness, LLM-as-judge rubrics, and merge gates on regressions.
Tracebench: Agent Trajectory Evals
Local-first harness for tool-using agents — golden traces, permission violations, and task-success scoring as offline evals.
Forge: Parallel Eval Runner (Rust)
High-throughput JSONL eval runner with bounded concurrency, deterministic replay, and CI-friendly exit codes.
Beacon: Live Eval Ops (Elixir)
Phoenix LiveView console for eval runs — OTP supervisors per job, PubSub progress, and a failure-review queue.
Crux API Layer
50+ provider aggregator for hotels, DMCs, and payments — 1M+ daily requests, sub-200ms hybrid search & RAG, 99.9% uptime.
Skills
AI & Agentic Systems
Agentic AI & MCP · GraphRAG & RAG · DSPy · vLLM / LangChain / LlamaIndex · AI Evals (PromptFoo, RAGAS, DeepEval, LLM-as-judge) · Agent traces & CI quality gates · LLMs (Claude, GPT-4, Gemini, Bedrock)
Backend & Languages
Node.js / TypeScript · Python · Go · Rust · Elixir / OTP / Phoenix · Java · SQL · Microservices
Cloud & Infra
AWS · Azure · Docker · Kubernetes · CI/CD · IAM / Security
Data
PostgreSQL / pgvector · MongoDB · Hybrid Search · GIS Pipelines
Education
- Chief Technology Officer Programme Indian Institute of Technology Delhi (IIT-D)
- Executive Programme in Product Management Indian School of Business (ISB)
- B.Tech in Computer Science & Engineering Bharati Vidyapeeth College of Engineering, New Delhi
Achievements
- 1st Position — Munjal University Hackathon (rapid prototyping & system design)
- Invited Mentor — HackBVP 2023 (scalable design, API contracts, agile)
HTML resume (indexable) · PDF version · Recruiter brief · llms.txt