● Open to 2026 opportunities

Hello. I'mAbhigyan Kumar Mahato

  • Artificial Intelligence & Machine Learning Engineer
  • Full Stack Developer
  • Problem Solver · Builder · Creator

I build intelligent systems and software that solve real-world problems through scalable engineering — from model research to production deployment.

View selected workGet in touch
neural core · interactive

01About

Engineering maturity over visual noise — systems designed to survive contact with real users.

Philosophy

Good engineering is mostly restraint. I optimise for legible architecture, measurable outcomes and interfaces that stay calm under load. Every model I ship comes with an evaluation story; every UI I build comes with a performance budget.

0+

Projects shipped

0

ML models in production

0

Hackathons & awards

0%

Uptime on deployed services

  1. 2026

    Applied AI Engineer

    Building retrieval-augmented systems and evaluation harnesses for production LLM features.

  2. 2025

    Full Stack Engineer

    Owned end-to-end delivery of an enterprise analytics dashboard used by 4k+ internal users.

  3. 2025

    ML Research

    Published work on advanced Traffic congession analysis and prediction using deep learning.

  4. 2024-2028

    B.Tech · Computer Science

    Specialised in machine learning, distributed systems and human-computer interaction.

02Skills

A constellation, not a progress bar. Hover any node to see how the stack connects.

Hover or tab through a node

Clusters

Also fluent in

  • Next.js
  • Supabase
  • GraphQL
  • Kubernetes
  • Weights & Biases
  • Hugging Face
  • Terraform
  • Playwright
  • pgvector
  • Grafana

Every cluster ships with the same discipline: typed contracts, measurable evaluation and observability from day one.

03Experience

Roles where the constraint was real: latency budgets, clinical data, production traffic.

AI/ML Engineer

2025 — Present

Independent · Applied AI

Design and ship retrieval, ranking and evaluation infrastructure for LLM-backed product features.

  • Cut hallucination rate 41% with a grounded retrieval + citation pipeline
  • Built an offline eval harness running 1.2k cases on every merge
  • Reduced inference cost per request by 3.4x through caching and routing
  • PyTorch
  • FastAPI
  • pgvector
  • Docker

Full Stack Developer

2024 — Present

Product Engineering

Owned an internal analytics platform end to end: schema, API, realtime layer and design system.

  • Shipped a virtualised grid holding 60fps at 100k rows
  • Moved p95 API latency from 1.1s to 240ms
  • Introduced typed contracts shared across four services
  • React
  • TypeScript
  • Node
  • Postgres

System view

How a request travels through the stack I build and operate.

  1. Frontend

    React · TypeScript · Design system

  2. Backend

    FastAPI · Node · Typed contracts

  3. Data

    Postgres · pgvector · Redis

  4. Cloud

    AWS · Docker · Terraform

  5. Delivery

    CI/CD · Observability · Rollbacks

04Selected work

Six builds that shipped, measured and survived real users.

05Credentials

Certifications, competitions and the numbers behind the practice.

Meta Full-Stack Developer

Meta

2025

AWS Certified Developer — Associate

Amazon Web Services

2025

OpenJS Node.js Application Developer

Linux Foundation

2025

MongoDB Certified Developer

MongoDB

2024

Deep Learning Specialization

DeepLearning.AI

2025

AWS Certified Solutions Architect

Amazon Web Services

2025

TensorFlow Developer

Google

2024

Machine Learning Engineering

Coursera

2024

8

Hackathons placed

1200+

DSA problems solved

3

Open source contributions merged upstream

2

Research write-ups

06Contact

Building something that needs both models and product sense? Let's talk.

Direct