adizen.ai adizen.ai Team Portfolio Get in touch

Software Engineer

Building scalable AI and backend systems with measurable impact.

I design and implement high-throughput backend systems, RAG pipelines, and production AI models for scale and performance.

  • Improved NER model macro F1 from 0.54 to 1.00 via gradient boosting.
  • Built streaming anomaly detection reaching 0.98 PR-AUC on chaos-injected telemetry.
  • Optimized GraphRAG pipelines to cut token usage by 42.8% versus vector-RAG.
Get in touch See selected work
Nilansh Jain

By the numbers

1+
Years
Professional experience designing and shipping products.
7
Projects
Projects highlighted in this portfolio.
21
Technologies
Tools and skills across the toolkit section.
150+
LeetCode problems Solved

The path so far

Journey

A short, honest arc — where I've built, what I owned, and what shipped.

June 2026Adizen.ai

Adizen.ai

Software Engineer

Optimization and architectural expansion of the platform, resolving core filtering issues and implementing an integration framework for shop-floor machinery.

January 2026 – May 2026ThreatX.ai

ThreatX.ai

Software Engineer Intern

Replaced regex taggers with 2-stage NER models and deployed standalone FastAPI microservices for hybrid retrieval.

September 2026 – November 2026CoderOne

CoderOne

Full Stack Intern

Led a 3-person team building a MERN chat platform with Redis pub/sub and implemented OAuth 2.0 with RBAC

What I ship at Adizen

Selected work

A few things I've built and shipped — swap in your own projects, screenshots, and outcomes.

Featured
Project

KisanCredit

Credit scoring for thin-file borrowers : a LightGBM default-risk model trained on 307K real loan applicants, served via FastAPI with SHAP explanations, a lender dashboard, and input-drift monitoring.

PythonFastAPILightGBMPyTorchSHAPPostgreSQL
Code ↗
Featured
Project

HELIOS

Real-time microservice anomaly detection with live SHAP attributions and Gemini-generated structured incident reports.

PythonGoKafkascikit-learnTimescaleDBPrometheus
Code ↗
Featured
Project

GraphRAG Pipeline Benchmark

Benchmarks three RAG pipelines: a plain LLM, vector RAG, and GraphRAG on a TigerGraph knowledge graph, for token cost versus accuracy. Built for the TigerGraph GraphRAG Inference Hackathon.

PythonFastAPITigerGraphPostgreSQLDockerReact
Code ↗
P3
Client work
JCD ERP Lite Custom Product Implementation

The JCD ERP Lite platform is an inventory and business logic management tool. The upgrade focused on streamlining production workflows and custom product management capabilities.

I led the optimization and architectural expansion of the platform, resolving core filtering issues and implementing an integration framework for shop-floor machinery.

  • Resolved critical filtering issues for designers and machinery.
  • Designed architecture for connected computer system integration.
  • Cleaned up redundant core application logic.
P4
Client work
YOS 2.0

YOS 2.0 provides sports health specialists with data infrastructure and analytics dashboards for patient management. It helps practitioners track health insights and improve clinical decision-making.

I delivered three scoped deliverables including API performance optimizations and custom correlation type representations.

  • Improved package and calendar fetch API performance.
  • Added support for custom correlation type representations.
Analytics & Dashboards

How I build

Toolkit

Languages · GitHub

Python JavaScript C# TypeScript

Machine Learning & AI

NLP Named Entity Recognition LLMs RAG Gradient Boosting Isolation Forest SHAP MLflow pgvector

Backend & Infrastructure

Python Go FastAPI

Ways of working

Iterative testing

Let's talk

Let's build something worth shipping.

nilanshjain0306@gmail.com LinkedIn · nilansh-jain-b32093303 GitHub · Nilanshjain
adizen.ai Built with the Adizen design system · Portfolio template v1