Department C · The Particulars

About,
in particular.

Born in Gujarat. Trained in computer engineering at PDEU. Have spent the last four years moving between data analyst, ML engineer, and full-stack engineer roles, usually because the project demanded it. Two IEEE papers. Three merged OSS patches. A long catalogue of finished things. Available for full-time and freelance.

Name
Jay Hemnani
Based
Gujarat, India, open to relocate
Status
Available, full-time and freelance
§ 01Curriculum vitae.5 roles
Summer 2025
full-time

Data Analyst

Elite Hotel Group · Remote
  • Engineered automated ETL pipelines using SQL and Python, reducing manual data preparation by 40% and improving data consistency across multi-property analytics.
  • Built interactive dashboards in Tableau and Power BI for occupancy, revenue, and KPI tracking across multiple hotel properties.
  • Developed demand forecasting models using time-series analysis to support dynamic pricing and revenue management decisions.
Python · SQL · Tableau · Power BI · Airflow
2022-2024
contract

Technical Consultant

Independent · Remote
  • Provided data analytics and pipeline consulting for small businesses, building reporting automation and data infrastructure solutions.
  • Designed A/B testing frameworks and tracked engagement metrics to optimize client campaigns and product decisions.
Python · SQL · Google Analytics · Mixpanel · Tableau · Figma
Jan-May 2022
internship

AI/ML Intern

Amnex · Gujarat, India
  • Built credit fraud detection system using ensemble ML (Random Forest, XGBoost) with SMOTE for class imbalance, achieving 94% precision on transaction data.
  • Developed automated analytics dashboards with statistical models for operational reporting and anomaly detection.
Python · Scikit-learn · TensorFlow · Pandas · NumPy · XGBoost
Jun-Aug 2021
internship

iOS App Development Intern

Cygnus SoftTech · Gujarat, India
  • Developed CodeLock, an iOS privacy/security app with AES encryption and secure local storage using Core Data.
  • Instrumented user engagement analytics and iterated on UX based on usage patterns and retention data.
Swift · UIKit · Core Data · Xcode · Git
May-Nov 2019
internship

SWE Intern

Cactus Creatives Pvt. Ltd. · Gujarat, India
  • Built cloud-native communication platform for first responders using microservices architecture on Azure with REST APIs.
  • Configured CI/CD pipelines for automated testing and deployment, reducing deployment time by 60%.
Azure · CI/CD · Microservices · REST APIs · Docker
Download the PDFsForward-DeployedData EngineerML EngineerBackend / SWEData Analyst
§ 02Education.one degree
2018-2022
GPA 8.7/10

B.Tech in Computer Engineering.

Pandit Deendayal Energy University (PDEU) · Gandhinagar, India

Four years of computer engineering, two IEEE papers as an undergrad, and a developing taste for systems that other people end up using.

CourseworkBig Data AnalyticsComputer VisionNatural Language ProcessingOperating SystemsDatabase SystemsSoftware Development
§ 03Publications.two IEEE papers
№ 01
IEEE International Conference on Machine Learning & Data Science·2022

Diabetes Prediction using Stacking Classifier.

Six-model stack achieving ~82.68% accuracy on PIMA dataset.

Paper ↗Code ↗
№ 02
IEEE International Conference·2021

CPU Scheduling Algorithms Analysis.

Comparative analysis of CPU scheduling algorithms with priority queue optimization.

Paper ↗Code ↗
§ 04Open source, three merged.vLLM · MCP SDK · A2UI

Three PRs into ecosystem-critical repositories, each landed by navigating an unfamiliar large codebase, finding the seam, and shipping the change.

PR #31513
vllm-project / vllm

Enable LoRA support for tower and connector in LLaVA.

Enabled LoRA training for the LLaVA vision tower and connector inside vLLM, the high-throughput LLM inference engine. Merged 2 January 2026.

View pull request
PR #1826
modelcontextprotocol / python-sdk

Add an explicit type annotation for the call_tool decorator.

Fixed a missing type annotation on the call_tool decorator in the Anthropic MCP Python SDK, the SDK that powers MCP-based integrations. Merged 6 January 2026.

View pull request
PR #407
google / A2UI

Add .env.example templates for safer secret setup.

Added env templates to Google's A2UI, the open standard for agent-to-user interfaces. A small patch with real surface area. Merged 5 January 2026.

View pull request
§ 05Stack, by honest level.expert · proficient · working

Expert ship daily, can design from scratch

  • Python
  • TypeScript
  • SQL
  • PyTorch
  • PostgreSQL
  • Docker
  • Kafka
  • Airflow
  • LangChain
  • LangGraph
  • MCP
  • Next.js
  • FastAPI
  • Node.js / Express

Proficient ship confidently, may reference docs

  • YOLOv8
  • ByteTrack
  • TimescaleDB
  • MongoDB
  • Redis
  • MLflow
  • DVC
  • Tableau
  • Power BI
  • AWS
  • Ollama
  • vLLM

Working shipped once or twice, ready to ramp

  • Go
  • Rust
  • Kubernetes
  • pgvector
  • Pinecone
  • Weights & Biases
  • GCP
  • Azure
  • Chrome Extensions
  • WebMCP
  • Three.js
  • C++
§ 06A few non-trivial particulars.numbered, not ranked
  1. 01

    Two IEEE publications, as an undergraduate.

    A research credential without a master's degree.

  2. 02

    Three merged OSS PRs into flagship repositories.

    vLLM, the Anthropic MCP SDK, and Google's A2UI. Evidence of landing changes inside unfamiliar large codebases.

  3. 03

    CAG Deep Research, built in ten days.

    Five-agent LangGraph system, hexagonal architecture, production-shipped. A signal for speed and ownership.

  4. 04

    Early WebMCP implementation in production.

    One of the first sites exposing structured tools via the W3C WebMCP standard. Eight tools, real adoption.

  5. 05

    Cursor, Claude Code, and Codex, daily for over a year.

    The honest version: faster ramps on unfamiliar codebases, tighter feedback loops, more time on the interesting work.

  6. 06

    Two production data platforms, two languages.

    Stock Data Platform in Python (Kafka, Airflow, TimescaleDB). Indian Stock Platform in Rust (Axum, WASM, Neon, GitHub Actions cron).

  7. 07

    WCA-registered speedcuber.

    Personal best of 16.7 seconds. Not on the resume, true regardless.

§ 07Get in touch.responses inside 48h

A note, an offer, or a problem worth solving.

I am looking for forward-deployed engineering, data engineering, agentic AI, and senior software roles. Staff-shaped problems welcome. India-remote, global-remote, and on-site with relocation all on the table.

Open the linejayhemnani992000@gmail.com
GitHub ↗LinkedIn ↗Twitter ↗Work →