Tribhangin Dichpally

Applied AI Engineer|Production ML & MLOps|

Available Summer 2026

Architecting deterministic agentic systems and high-throughput data pipelines. Focused on Bridging the gap between LLM research and production scale.

About Me.

I build Applied AI systems that don't just predict, but perform. My focus is on creating deterministic orchestration patterns that make LLMs reliable in production environments.

Expertise

Technical Focus.

Portfolio

Featured Projects.

PythonLangGraphSQLiteAnthropicGeminiFastAPI

Market AI Portfolio Pipeline

5-Stage Autonomous Multi-Agent Equity Research System

Architected LangGraph adversarial debate engine across growth, value, macro, and risk specialist agents producing per-stock conviction scores and bull/base/bear scenario models.

Built ETL research-memory layer ingesting filings, exchange data, news, and analyst PDFs into chunked SQLite-backed document packets with per-symbol memory.

Shipped dynamic LLM provider routing across Anthropic, Gemini, Groq, Mistral, and OpenAI-compatible endpoints with per-stage model config for cost and latency optimization.

pipeline5-stage

agents4 specialist

modeAutonomous daemon

ReactTypeScriptNode.jsExpressGeminiAnthropicSSE

Code Review Agent

Agentic GitHub PR Review with Human-in-the-Loop

Architected multi-turn agentic review loop using Gemini with Anthropic Claude fallback, reading full file context and PR diffs via tool calls and emitting structured comments across 5 severity-tagged categories.

Designed React + TypeScript review dashboard consuming SSE streams to render inline annotations in real time, with human-in-the-loop approve/dismiss controls gating all agent comments before posting to GitHub.

Shipped GitHub OAuth 2.0 flow with repo/PR picker, webhook auto-registration, and session persistence enabling zero-config onboarding for new repositories.

categories5 severity

file types15+

routingMulti-LLM

FastAPIQdrantRedisReactTypeScriptGroq

DocuGenie

Agentic RAG Application with Streaming UI

Built React + TypeScript conversational interface with incremental token streaming and ReAct-style agent orchestration enabling dynamic retrieval and reranking based on query intent.

Reduced end-to-end latency from 1.2s to <0.7s (40% improvement) via Redis-backed semantic caching; cut LLM API costs by 30% via optimized inference call patterns.

latency< 0.7s

cost-30%

improvement40%

Scholarly Work

Research & Publications.

2023

Advancements in Hindi-English Neural Machine Translation: Leveraging LSTM with Attention Mechanism

International Research Journal of Engineering and Technology (IRJET)

Leveraging LSTM with Attention Mechanism for improved translation quality between Hindi and English.

Neural Machine TranslationAttention Mechanism

2025

A Comparative Analysis of Deep Learning Models for Waste Segregation

Multimedia Tools & Applications (Springer)

A deep dive into deep learning architectures for automated waste classification and segregation.

Springer PublicationModel benchmark

2024

Feature Extraction and Diagnosis of Heart Diseases Using Fuzzy‑Based IoMT

Advances in Fuzzy‐Based IoMT (Wiley)

Privacy-preserving diagnosis framework for heart diseases using fuzzy logic and Internet of Medical Things.

Wiley PublicationFuzzy-Based IoMT

Get in touch

Build Together.

Drop a message

Interested in collaboration or have a question about my work? Feel free to reach out. I'm currently open to Applied AI / MLOps roles for Summer 2026.

Ask my AIRAG Powered

Hi! I'm Tribhangin's AI assistant. Ask me anything about his projects, research, or experience!