Devanshu Chicholikar
Software Engineer · AI Engineer · Forward Deployed Engineer · Boston, MA (Open to Relocation)
chicholikar.d@northeastern.edugithub.com/DevanshuNEUlinkedin.com/in/devanshuchicholikarDownload PDF
AI engineer who ships AI to production end to end: MCP servers, RAG with retrieval evals, LLM-as-a-judge eval harnesses and voice agents, plus the full-stack, infra and design work around them. OpenCodeIntel is a code-search platform for AI coding agents (web app, REST API, 12-tool MCP server) at 94% Hit@1 on a 14-codebase research eval; Overhear grades voice-agent calls with code checks plus an LLM judge on a 39-call golden dataset; CallBudget cuts expected pharmacy calls 47% with a voice agent that abstains instead of guessing. Earlier: two years full-time on a Java / Spring Boot quotation platform, and teaching cloud computing on AWS to 100+ graduate students.
Graduate Teaching Assistant, CSYE 6225 Network Structures and Cloud Computing
Northeastern University · Sep 2025 - May 2026 · Boston, MA
- Taught AWS, Terraform, and distributed systems to 100+ graduate students across two semesters; authored a Docker + GitHub Actions CI/CD lab adopted as official course content across 3 sections (180+ students)
- Led system-design and code reviews for 15 cloud-native API project teams, coaching on scalability, fault tolerance, and API design
Software Engineer, B2B Industrial-Equipment Platform
Jaksh Enterprise · Aug 2022 - Jul 2024 · Ahmedabad, India
- Developed a Java / Spring Boot product-catalog and quotation engine for 590+ customizable products serving a 10K+ B2B customer base, over PostgreSQL and a rules-based pricing module
- Cut quote-page p95 latency 65% (800ms to 280ms) via PostgreSQL indexing, Redis caching, and async processing
- Owned delivery end-to-end as one of two engineers: gathered client requirements, scoped functional specs, and shipped via Dockerized CI/CD on GitHub Actions, cutting release cycles from 2 weeks to 3 days
Software Development Engineer Intern
Pitney Bowes · Jan 2022 - Jul 2022 · Pune, India
- Built REST APIs for PitneyShipPro with idempotent endpoints, eliminating 40% of manual data entry; raised automated test coverage to 85% with Jest and Cypress
OpenCodeIntel · Nov 2025 - Present
Python · FastAPI · FastMCP · tree-sitter · Pinecone · Redis · React · TypeScript
Code-search platform for AI coding agents: a web app, a REST API and an MCP server exposing 12 tools (semantic code search, dependency graph, impact analysis, context assembly) over stdio and streamable HTTP. Benchmarked retrieval to 94% average Hit@1 across 14 OSS codebases (665-query research eval), +8.4 points from reranking (Voyage rerank-2) isolated via a 98-run ablation. Public playground search: p50 641ms cold, 242ms cached.
Overhear · Sep 2026
TypeScript · Next.js · Retell · Claude API · Postgres · Drizzle · Vitest
AI QA analyst for voice agents: grades every call a Retell scheduling agent takes on a 7-dimension rubric. 3 dimensions (50% of the score) are decided in code from the call's action log; an LLM-as-a-judge checks what the agent said against the real open slots and scores the rest. Caught 23 of 23 planted failures on a 39-call golden dataset, macro-F1 0.89, 0.99 self-consistency across 3 runs.
CallBudget · 2026
Python · FastMCP · scikit-learn · Optuna · Pipecat · Deepgram · DuckDB
Agentic pharmacy-stock search: a FastMCP server (predict / plan / eval / converse) over a learned stock-probability ranker cuts expected calls-to-find from 4.3 to 2.3 (47%) on a 19-pharmacy simulation. An abstention layer on the stock extractor (self-consistency voting) drives false "in stock" answers from 10% to 0% under a deliberately unreliable extractor; a Claude-driven voice agent, tested against a simulated pharmacist, handles the calls.
Saar · Mar 2026 - Present
TypeScript · WXT · Chrome MV3 · React · Vitest · Playwright
Chrome MV3 extension on the Chrome Web Store that tracks Claude.ai token usage and cost in real time, entirely client-side: a page-context interceptor decodes Anthropic SSE streams, a Shadow-DOM overlay renders usage, and a service worker runs a BPE tokenizer. 1,808 Vitest tests across 63 files.
SecureScale · 2025
AWS · Terraform · Packer · Docker · Lambda · RDS · KMS · CloudWatch
Multi-AZ AWS infrastructure (VPC, ALB, ASG, RDS, S3 with KMS) as modular Terraform with IAM least-privilege security, cutting provisioning from 2 hours to 10 minutes and cloud spend 30% at 99.9% uptime.
M.S. in Software Engineering Systems
Northeastern University · Sep 2024 - May 2026 · Boston, MA
GPA: 3.85 · Generative AI · MLOps · Distributed Systems · Cloud Computing · Database Management Design · Algorithms
B.Tech in Information and Communication Technology
Dhirubhai Ambani Institute of ICT · Aug 2018 - May 2022 · Gujarat, India
Computer Networks · Operating Systems · Data Structures · Databases
- AI-Assisted Development
- Claude Code (subagents, hooks, custom skills), MCP server development, Agentic workflows, Spec-first development with AI agents, LLM evals, Cursor
- AI & Agents / Eval
- Agent tool-calling, RAG (hybrid AST + embeddings, BM25 + reranking), LLM evaluation (LLM-as-a-judge, golden datasets, calibrated abstention, self-consistency, guardrails), Voice agents (Pipecat, Deepgram), tree-sitter, Context engineering
- Languages
- Python, TypeScript, Java, JavaScript, Go, SQL
- Backend & APIs
- FastAPI, Spring Boot, Node.js, REST, gRPC, WebSocket, React, Next.js
- Cloud & DevOps
- AWS, GCP, Docker, Kubernetes, Terraform, GitHub Actions, CI/CD, Vercel, Railway
- Databases
- PostgreSQL, Redis, Pinecone, DuckDB, Supabase, MongoDB
- Foundations
- Data structures & algorithms, Distributed systems, OOP & design patterns, System design