Open to AI/ML Opportunities • Based in India

Building intelligent systems.

Computer Science Engineering student building AI-powered products across Generative AI, NLP, healthcare technology, and full-stack software foundations.

Explore My Work arrow_downward mail Email Me GitHub ↗

DEV_NODE // 01

AI BUILDER

LAT 28.6139° N · LON 77.2090° E

Sudhanshu Shekhar Jha

AI Builder • Student Engineer

BUILD-MODE // ONLINE

memory AI / SOFTWARE STACK

Python FastAPI Flask React 19 Next.js Groq / Llama 3.3 PostgreSQL HL7 FHIR R4 Gemini API Docker

/ 01 — Philosophy

I don't just build AI demos.
I build systems around real problems.

Good engineering begins with understanding the problem—not choosing a framework. I use AI where it creates meaningful leverage, combine it with reliable software architecture, and focus on building products people can actually use.

01. EMPATHY FIRST

Human-Centric Problems

Targeting anxiety detection, patient queue bottlenecks, and clinical overhead.

02. LATENCY DISCIPLINE

Sub-Second Inference

Leveraging Groq LPUs and optimized token serialization for fluid experiences.

03. STANDARDS DRIVEN

Interoperability

Strict adherence to HL7 FHIR R4 specifications, clean SQL schemas, and resilient microservices.

PERFORMANCE CORE

LOW-LATENCY
AI APPLICATIONS

REAL-TIME INTERACTION

Architecture Focus Low-Latency Systems

DEVELOPMENT CREED

“Build. Learn. Experiment. Ship. Repeat.”

— SUDHANSHU SHEKHAR JHA

/ 02 — About

Computer Science Engineering student building at the intersection of AI and software engineering.

neurology

01 AI / ML

Generative AI, NLP & Multimodal

Hands-on implementation of LLM reasoning loops, speech-to-intent pipelines, contextual retrieval, and multimodal document understanding.

Generative AI · NLP · Multimodal Systems

code_blocks

02 ENGINEERING

Full-Stack Foundations

Constructing reliable web backends, responsive user experiences, clean relational schemas, and asynchronous microservices.

Python · FastAPI · Flask · React · Next.js · SQL

rocket_launch

03 BUILDING

Velocity & Prototypes

Iterating fast through national hackathons, open-source tooling, and shipping working prototypes with live users in mind.

Hackathons · Open-source · Production AI Prototypes

/ 03 — Selected Work

Featured projects.

Built from first principles with verified live deployments and open codebases.

PROJECT // 01 FEATURED PROJECT • LIVE DEMO

Student Wellness AI

An AI-driven student wellness platform designed to provide accessible conversational support, identify potentially urgent distress signals, support guided decompression activities, and maintain structured wellness interactions.

USER INPUT ─→ LLAMA 3.3 ─→ SIGNAL TRIAGE ─→ DECOMPRESSION

AI CONVERSATION DISTRESS SIGNALS WELLNESS TRACKING LIGHTWEIGHT DEPLOYMENT

Llama 3.3 Groq API NLP Flask SQLAlchemy Tailwind CSS

Live Demo ↗ code GitHub ↗

lock student-wellness-app.onrender.com

ACTIVE

Evening Check-In · Guided Session Active Mindfulness Triage

State: Academic Overwhelm → Decompression Phase Safe Space

SESSION: GROQ // LLAMA-3.3-70B LATENCY: ~180ms

Student

I'm feeling overwhelmed with exam prep and upcoming project deadlines...

smart_toy AI Assistant

Let's break this down into manageable blocks. Take a slow breath—what's the single highest-priority task right now?

vital_signs CLINICAL_STREAM // MULTIMODAL INTAKE

ABDM ALIGNED

VOICE

OCR

EXTRACTION

FHIR R4

OPD QUEUE

AUDIO STREAM WEB SPEECH

"Acute dry cough and mild fever for 3 days..."

OCR SCAN ENGINE LAYOUTLM

Prescription slip extracted • 2 items detected

// FHIR R4 Bundle preview snippet TRANSACTION BUNDLE

{
  "resourceType": "Bundle",
  "type": "transaction",
  "entry": [
    {
      "resource": {
        "resourceType": "Condition",
        "code": "Cough",
        "clinicalStatus": "active"
      }
    }
  ]
}

Target OPD: Internal Medicine Structured for OPD Queue • ABDM Aligned

PROJECT // 02 FLAGSHIP HEALTHCARE • ACTIVE BUILD

AI Clinical Intake Platform

Multimodal clinical intake system designed to reduce intake overhead by structuring patient speech and prescription document inputs into ABDM-aligned HL7 FHIR R4 clinical packages before doctor consultation.

VOICE-FIRST INTAKE DOCUMENT OCR STRUCTURED DATA FHIR R4

Gemini API Web Speech API OCR / LayoutLM FastAPI React HL7 FHIR R4 ABDM

Code Base ↗ account_tree Architecture Spec ↗

/ 04 — Capabilities

What I build with.

Building practical AI applications with modern models, reliable APIs, and scalable software foundations.

neurology

01 Generative AI Systems

LLM orchestration, structured prompt engineering, Groq & Ollama Llama inference, and Gemini API integration.

LLAMA 3.3 • GROQ • GEMINI

psychology

02 NLP & Language Pipelines

Intent detection, sentiment tracking, contextual embeddings, semantic search, and modular RAG workflows.

INTENT • EMBEDDINGS • RAG

graphic_eq

03 Multimodal AI

Audio stream processing via Web Speech, prescription document OCR extraction, and vision-language integration.

AUDIO • OCR • VISION

terminal

04 Full-Stack Applications

Asynchronous FastAPI & Flask backends, React & Next.js interfaces, PostgreSQL / SQLAlchemy, and Docker deployment.

FASTAPI • REACT • POSTGRES

TECHNOLOGY ARSENAL

AI / SOFTWARE STACK

Python React Next.js FastAPI Flask Groq / Llama 3.3 Gemini API PostgreSQL SQLAlchemy Docker HL7 FHIR R4 Git

/ 05 — The Journey

Learning by shipping.

Milestones, project deployments, and the technical roadmap forward.

2026 // CURRENT — AI PRODUCT DEVELOPMENT LIVE DEMO

Built and deployed full-stack AI prototypes

Deployed full-stack end-to-end applications to solve psychological wellness telemetry for university students (student-wellness-app), and created Medkiosk to map real-time OPD hospital triage data into standard HL7 FHIR protocols.

LLAMA 3.3 • GROQ • FASTAPI • FHIR R4

ONGOING — HACKATHONS & REAL-WORLD EXPERIMENTS ACTIVE BUILD

Transforming raw ideas into working software under rapid hackathon deadlines

Consistently competing in collegiate and national engineering hackathons. Focusing on reliable API integrations, resilient end-to-end user workflows, and sub-second delivery.

NEXT HORIZON — SCALABLE PRODUCTION AI ROADMAP

Resilient multi-agent orchestration & modular RAG

Researching resilient multi-agent orchestration, modular Retrieval-Augmented Generation (RAG) with vector search, localized edge inference, and automated evaluation suites.

RESEARCH & PROTOTYPE HORIZON

Currently Exploring

Active technical investigations

AGENTIC AI

RAG & VECTOR

MULTIMODAL

AI EVALUATION

AI SAFETY

OBSERVABILITY

/ 06 — Contact

Let's build something meaningful.

Have an interesting problem, a project worth collaborating on, or an opportunity for an ambitious AI builder? Let's connect.

DIRECT COMMUNICATION CHANNEL

sudhanshushekharjha2007@gmail.com

outgoing_mail Email Directly GitHub ↗ LinkedIn ↗

MESSAGE SYNTHESIZER

Initiate Connection

READY