Skip to content
View KshitijAng's full-sized avatar
2 followers · 0 following

Block or report KshitijAng

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
KshitijAng/README.md

Hi, I'm Kshitij 👋

AI engineer building backend systems and LLM pipelines. Currently shipping multi-phase LLM workflows, AI adoption analytics, and chat-agent automation at Hiver — an AI-powered customer service platform.

Outside work, I build production-style side projects in backend + AI infrastructure to feel the rough edges of modern AI engineering hands-on.

Experience

Hiver — AI Intern · Dec 2025 – Jun 2026

  • Owned key AI Adoption/ROI initiatives across backend and frontend, leveraging Redis for precomputed metrics, integrating AI workflows powered by Elasticsearch, Gainsight, and MySQL, and shifting heavy aggregations to backend services.
  • Architected a 5-phase AI Tasks Opportunity Pipeline, running 21 cookbook recipes across 5 batched LLM (GPT-5.4) calls, with retry backoff on 429s and rate-limit pacing; added signal-keyword evidence filtering for accuracy.
  • Built an end-to-end AI automation testing platform for Chat Agents (React + FastAPI + Playwright), enabling automated execution of a 15-step happy-path scenario and per-run artifact lifecycle (replay video + failure screenshots).
  • Designed and implemented a daily AI adoption pipeline that computes per-UG daily AI usage metrics separate from 30/60/90-day windows and publishes feature-wise JSON snapshots to S3 using SQS-based orchestration and batched processing.

Ernst & Young — Generative AI Intern · Sep 2024 – Nov 2024

  • Reduced LLM setup friction by scripting model provisioning and standardizing config across Llama 3.1 deployments, gaining hands-on exposure to Transformer architecture and open-source LLM tooling.
  • Established a RAG-based Gen AI solution using LangChain, OpenAI, and Pinecone; tuned chunking and retrieval parameters to deliver contextually relevant responses on internal document queries.

What I've built

🔍 HybridRAG — Hybrid-search RAG over technical documentation (pgvector + Postgres FTS + RRF + Cohere rerank-3.5) with citation-grounded answers and Langfuse traces. Article writeup ↗

🎫 TicketSense — LLM-powered support ticket triage with Pydantic structured outputs, two-tier storage (Postgres + Redis), and a Groq primary + fallback model setup. Processed 300 tickets in ~12 minutes with zero failures.

🧠 AI Hub — Spring Boot 3.5 + React 19 chat & code-generation app on Spring AI + Groq, with reusable PromptTemplate definitions shared across routes.

📦 FlowVentry — Next.js 15 + MongoDB inventory app with cached MongoClient singleton and $text aggregation pipeline for relevance search.

Stack I work with

  • Languages: Python · Java · JavaScript · TypeScript · SQL
  • AI / LLM: LangChain · Langfuse · Pydantic · Cohere · Groq · OpenAI
  • Backend: FastAPI · Spring Boot · Spring AI · Next.js
  • Data: PostgreSQL · pgvector · Redis · MongoDB · Elasticsearch
  • Infra: Docker · Kubernetes · AWS · Kafka

Connect

LinkedIn · kshitijangurala903@gmail.com

Open to backend / AI engineering opportunities — Bengaluru, remote, or hybrid.

What I've been exploring lately

  • Hybrid-search retrieval combining sparse + dense fusion (RRF) and cross-encoder reranking
  • Citation-grounded LLM answers — making generation traceable to source chunks
  • End-to-end LLM observability with Langfuse traces across retrieval and generation
  • Multi-phase LLM pipelines with batched API calls, retry backoff, and rate-limit pacing
  • AI automation testing for Chat Agents using React + FastAPI + Playwright
  • AI adoption analytics pipelines built on SQS-based orchestration with feature-wise S3 snapshots

Pinned Loading

  1. Hybrid-RAG-with-pgvector-Cohere-Rerank-3.5 Hybrid-RAG-with-pgvector-Cohere-Rerank-3.5 Public

    Hybrid-search RAG over technical docs - pgvector + Postgres FTS fused via Reciprocal Rank Fusion, Cohere rerank-3.5, citation-grounded answers, Langfuse traces.

    Python

  2. TicketSense TicketSense Public

    LLM-powered support ticket triage. Groq with primary + fallback models, Pydantic structured outputs, Postgres + Redis two-tier storage, Streamlit dashboard.

    Python

  3. FlowVentry-Inventory-Management FlowVentry-Inventory-Management Public

    Next.js 15 + MongoDB inventory app - module-level cached MongoClient singleton, $text aggregation pipeline with a compound text index for relevance search.

    JavaScript

  4. AI-Hub-SpringBoot AI-Hub-SpringBoot Public

    Spring Boot 3.5 + React 19 chat + code-generation app on Spring AI and Groq's OpenAI-compatible API, with reusable PromptTemplate definitions across routes.

    Java

  5. Churn-Prediction-Decision-Tree Churn-Prediction-Decision-Tree Public

    Class-balanced decision-tree churn classifier (one-hot, CV-tuned max_depth) - lifted churn-class recall 0.43 → 0.77 over a 4-feature baseline. FastAPI + ONNX.

    Jupyter Notebook