Pushkal Pratap Singh | AI/ML & Backend Engineer

01. About Me

I'm an AI/ML & Backend Engineer who loves shipping LLM gateways, RAG pipelines, and production-grade ML microservices.

My work lives at the intersection of applied machine learning and solid backend engineering — fine-tuning transformer models, architecting OpenAI-compatible gateways with intent-aware routing, and orchestrating observable, containerized services with FastAPI, Postgres, Redis, and vector databases.

I'm pursuing my B.Tech in Computer Science at SRM Institute of Science and Technology (Sem 1–6 CGPA 9.06). Recently a Smart India Hackathon 2025 National Finalist and AI For Bharat (AWS) 2nd Runner-Up.

9.06 CGPA

35+ Repositories

2+ Internships

7 Models Fine-tuned

Download Resume

02. Skills & Technologies

CS Fundamentals

Data Structures Algorithms Operating Systems DBMS System Design

ML / AI

LangChain RAG Pipelines Fine-tuning Hugging Face PyTorch spaCy LLM Gateways Prompt Engineering

Vector Stores

Pinecone FAISS Chroma pgvector

Backend & Languages

Python SQL FastAPI Pydantic REST APIs PostgreSQL Redis MongoDB

Cloud & DevOps

AWS Lambda API Gateway DynamoDB SAM Azure Docker Docker Compose CI/CD

Observability

OpenTelemetry Prometheus Grafana Kafka

03. Experience

Dec 2025 - May 2026

Backend Developer Intern

Tranquiex · Remote

Developed and maintained REST API endpoints and database schemas supporting core product functionality in production
Partnered with frontend engineers on feature delivery, bug triage, and code review across sprint cycles

Jan 2025 - Dec 2025

AI Training Contractor

Outlier AI · Remote

Evaluated LLM outputs across coding, reasoning, and math domains; authored prompt-response pairs and rubric-based assessments to improve model alignment
Performed prompt optimization and dataset quality audits, flagging hallucinations and ambiguous specifications across hundreds of task batches

2024 - 2025

President

SRMIST RMP

Led strategic initiatives, mentored teams, and drove innovation across events and tech projects
Coordinated hackathons, workshops, and community outreach programs

04. Featured Projects

35+ Repositories on GitHub

View all on GitHub

⭐ Flagship

Prompt Intelligence Layer (PIL)

An opinionated, OpenAI-compatible LLM gateway with intent-aware routing across OpenAI, Anthropic, Gemini, and local Ollama behind a unified /v1/chat/completions API. Features a semantic cache (Redis + pgvector), per-key budgets, circuit breakers, PII masking with reversible de-tokenization, provider fallback chains, and end-to-end OpenTelemetry tracing with shadow-mode replay for safe rollouts.

FastAPI Postgres + pgvector Redis OpenTelemetry Presidio LLM Gateway

Python

🔧 Serverless AI

ZIT — Git Workflow Automation

A serverless AI backend on AWS Lambda + API Gateway with a dual-model strategy (Claude 3 Sonnet for complex reasoning, Claude 3 Haiku for lightweight requests) to cut token cost while preserving quality. Deployed via AWS SAM/CloudFormation with scoped IAM and 27 pytest cases, plus a RAG pipeline using FAISS for vector search and DynamoDB as the metadata store.

AWS Lambda API Gateway Claude 3 FAISS DynamoDB AWS SAM

Python

⭐ Featured

Nyay Samiti

A legal document intelligence platform built on 7 fine-tuned transformers (LegalBERT, RoBERTa, BART, T5, Sentence-BERT) over Indian legal corpora — 90.4% F1 (NER), 84.16% accuracy (Risk Scoring), and 87.18% Exact Match (Q&A). A FastAPI ML microservice runs a 7-stage inference pipeline (~15s CPU / ~3s GPU), orchestrated as a 6-service stack with Kafka, Prometheus, and Grafana via Docker Compose.

LegalBERT PyTorch spaCy FastAPI Kafka Grafana

Python

Cortex

A brain for your company — a centralized knowledge and intelligence layer that ingests organizational data and exposes it through retrieval and reasoning, so teams can query, surface, and act on institutional knowledge in one place.

Python LLM RAG Knowledge Base

Python

Backtest Dashboard

A high-performance dashboard to backtest trading logic and strategies at scale, built in Rust for speed and reliability. Run strategy simulations and analyze performance metrics across large historical datasets.

Rust Backtesting Quant Dashboard

Rust

1

CodeGraph

A code intelligence tool that maps a codebase into a navigable graph of files, symbols, and dependencies — surfacing structure and relationships to make large repositories easier to understand and reason about.

Python Static Analysis Graphs AST

Python

04.5 GitHub Activity

@xpushkal

-- repos -- followers -- following

-- contributions in the last year

Loading contributions...

05. Education

2023 - 2027

SRM Institute of Science and Technology

Bachelor of Technology, Computer Science

Ramapuram, Chennai

CGPA: 9.06 / 10 (Sem 1–6)

2022 - 2023

City Montessori School

12th Grade - PCM & Computer Science

Lucknow

90.2%

2019 - 2020

City Montessori School

10th Grade - Science & Computer

Lucknow

92.5%

06. Achievements

🏆

Smart India Hackathon 2025 — National Finalist

Among the top teams selected from 50,000+ registrations nationwide.

🥉

AI For Bharat 2026 (powered by AWS) — 2nd Runner-Up

Placed among the top teams out of 96,000+ developers.

🎯

DEVFEST 2025

Attended DEVFEST organized by GDG Chennai at IIT Madras Research Park.

🚀

IIM Bangalore — NSRCEL SummitUp 2024

Attended the startup summit to understand the technical and entrepreneurial industry.

07. Get In Touch

I'm always open to discussing new projects, creative ideas, or opportunities to be part of your vision. Feel free to reach out!

pushkalx30@gmail.com +91 8756234999