Umarfarook Gurramkonda

Folio, Edition 2026. Seven chapters.

Bengaluru, India (UTC+5:30)

GitHubLinkedInCal.com

About me

Folio · Edition, the author

Hi, I am Umarfarook, a founding ML engineer with 2+ years in LLM systems, NL-to-SQL and the web apps around them, shipping features behind an eval harness and a hard cost cap.

I measure what I ship.

The day job

what I spend the day on

I decide when a model earns its place, then check the output still holds under real traffic. Most of the work is evaluation and cost discipline. Neither is fun. I want applied ML and LLM engineering roles at early-stage companies, where that judgment counts as much as model choice.

Career journey

2020 to now

  • Oct 2024 to nowFounding ML EngineerHypeon AI

    I own the LLM orchestration service: routing, intent and composition sub-agents streaming over SSE, plus NL-to-SQL over BigQuery that dry-runs every query before it costs anything. Claude Haiku is primary, Gemini is the fallback.

    LangChain · FastAPI · BigQuery · Cloud Run · Claude · Gemini

  • Jun to Sep 2024Backend Developer InternSynclovis Systems

    I built REST services in Node, Express and MySQL for an event-management app. I also added LangChain and FAISS retrieval to an internal LLM health assistant, with guardrails that refuse out-of-scope questions.

    Node · Express · MySQL · LangChain · FAISS

  • 2020 to 2024B.Tech · Computer Science · 2024K.S.R.M College of Engineering

    K.S.R.M College of Engineering · 8.14 / 10

Certifications

Oracle, Azure and AWS

  • Oracle OCI Data Science Professional · 2025
  • Oracle OCI AI Foundations Associate · 2025
  • Azure AI Fundamentals · AWS Cloud Foundations

Stack

Everything here is in something I shipped.

Models + agents

  • Claude
  • Gemini
  • OpenAI
  • LangChain
  • sentence-transformers
  • Hugging Face

Backend + data

  • Python
  • FastAPI
  • PostgreSQL
  • BigQuery
  • FAISS
  • Redis
  • Pydantic

Delivery

  • GCP
  • AWS
  • Docker
  • GitHub Actions
  • Alembic
  • Prometheus

Interface

  • TypeScript
  • Next.js (React)
  • Tailwind CSS
  • SSE

Method

Same four stages on every project.

  1. 01OrchestrateCargo Concierge ↗

    I route each request through extraction, rate lookup, ranking and drafting, and stream progress while they run. One quote fans out to six model calls: extraction, up to three rationales, a recommendation and the draft email. Extraction averages 2,680 ms in the 15-item ablation. The end-to-end quote is not timed.

  2. 02GroundRAG Document QA · v0.0.1 ↗

    I put the vector store behind a Protocol so the backend is a one-file swap, and the retrieval metrics (Recall@K, MRR, nDCG@10) are a component with a CLI rather than a script run once. The harness has not been pointed at a benchmark yet. Below a confidence floor the answer comes back as “I don’t know” plus the closest passage. Citations are required or the answer is rejected.

  3. 03Constrainmcp-bigquery-evals ↗

    I dry-run every BigQuery call and refuse anything that would scan past 100 MB. Seven read-only tools, seven stable error codes an agent can switch on. There is no write path.

  4. 04MeasureTrustBench ↗

    I score every answer on eight metrics, five judged and three deterministic. When a slice regresses I run McNemar’s exact test before I call it real. The Cohen’s kappa function for judge-versus-human calibration is written and tested, but I have not labelled a set to run it against. 82 tests run offline with no API key.

Open for work

Bengaluru, India (UTC+5:30)

available now, 15-day notice, remote or contract. I work US, EU or AU hours from Bengaluru.

Resume ↗