AI/ML Engineer — Chandler, AZ2026

Years of experience

Projects built

+0K

Images processed in prod

HelloI'm Sravanakumar Sathish

—I build AI systems that ship: production RAG, agentic workflows, computer vision, and LLM-powered products.

2.5+ years across PwC, Honeywell Aerospace, qBotica, DriverAI, and Atsuya. M.S. in Data Science at ASU, graduating 2026.

Open to full-time rolesMay 2026 startF-1 OPT · STEM eligible

See what I've built View resume Download

tap the bot to open…

Available for work

Scroll down

Why hire me

Production, not prototypes

PwC, Honeywell Aerospace, qBotica, DriverAI — systems that ran for real users with real constraints, not demo repos.

Safety-engineered AI

Validation gates against hallucinated prices, evidence-cited clinical answers, and client-side failover when the backend dies.

Full-stack ML

Fine-tuning and quantization through FastAPI services to React front ends and cloud deployment on AWS, Azure, and GCP.

Agentic AI · RAG · LLM fine-tuning · LangChain · LangGraph · PyTorch · FastAPI · MLOps · AWS · Azure · GCP

01 — Featured work

Three things I'm proud of

Each one shipped against a real constraint: hallucination risk in clinical answers, downtime in fleet dispatch, and invented prices at checkout. Click any card for the full story.

Dec 2025 – Feb 2026

RACE

RAG-Optimized Clinical Reasoning Engine

Clinical language models hallucinate facts that can endanger patients, and the models capable of real reasoning are too large to run anywhere near the point of care.

Skills used

PythonLlama-3-8BQLoRARAGLangChainChromaDBFAISSPyTorch

I take models from research notebooks to constrained hardware — and I engineer against hallucination as a first-class risk, not an afterthought.

Read the case study View code

Impact

5.5 GB: Compressed model size
<1s: Evidence retrieval latency
8 GB: VRAM deployment target met
0.1%: Parameters trained via QLoRA

Mar 2026 – Present

TruckerPath

Fleet Operations & Intelligence Platform

Fleet teams juggle dispatch, HOS compliance, and backhaul procurement across disconnected tools — and a single backend outage stops load matching entirely.

Skills used

React 19VitePythonFastAPIThree.jsreact-three-fiberSQLitePydantic v2

I design for failure: when the backend goes down, this platform degrades gracefully to client-side compute instead of stopping the business.

Read the case study View code

Impact

4-gate: Carrier eligibility pipeline
<2s: Failover to client-side matching
280+: US cities in haversine routing
3D: Live component-level fault telemetry

Mar 2026 – Present

BayOps AI

Voice-Driven Parts Ordering for Auto Repair Shops

Service advisors lose hours on multi-vendor parts ordering, and LLM-driven checkout flows quietly invent prices and order parts that don't fit the vehicle.

Skills used

PythonFastAPIReact 19GroqElevenLabsLiveKitTwilioPlaywright

I don't trust LLM output by default — every money-touching step passes a deterministic validation gate before it hits a real cart.

Read the case study View code

Impact

0: Hallucinated-price checkout errors
Multi-vendor: Ordering from one voice flow
Real-time: Fitment verification via vPIC
Live: Excel billing sync

02 — Everything else

18 more things I've built

Multi-agent systems, vision models, edge inference, and full-stack platforms — including two hackathon wins and an IEEE publication. Filter by what you're hiring for.

Hiring for:

All repos on GitHub

InnovationHacks 2.0Apr 2026

Pipeline

Voyager AI

Agentic travel planning platform on Next.js and Vercel with topic scoping, hallucination prevention, and structured itinerary output across the full agent workflow.

OAuth 2.0: Calendar-grounded availability
6-stage: Guided planning workflow

Pipeline

Voice intake

Calendar OAuth

Gemini planning

ElevenLabs response

Vercel deploy

View project

Multi-Modal Vision-Language Model for Cancer Diagnosis project visual

ResearchJan – Aug 2025

Pipeline

Multi-Modal Vision-Language Model for Cancer Diagnosis

Cross-modal pipeline correlating high-dimensional medical imaging with unstructured biomedical literature, with a Grad-CAM + Gradio explainability interface for clinical interpretability.

ImpactBefore → After

Text-only reports→80%: Validation accuracy on MRI data
Black-box→Grad-CAM: Per-prediction explainability

Pipeline

MRI imaging

Vision Transformer

BioBERT fusion

Grad-CAM

Gradio UI

View project

Multi-AgentDec 2025 – Jan 2026

Pipeline

Zone Scout

Multi-agent lead intelligence system with an automated enrichment layer over Google Maps Places API — verified contacts, reviews, business metadata, and social profiles per validated lead.

ImpactBefore → After

1-2→4+: Enrichment sources per lead
Manual→Zero-touch: Lead validation pipeline

Pipeline

Map screenshot/ZIP

Gemini geocoding

Llama audit

Places enrichment

Streamlit UI

View project

Multi-AgentOct 2025 – Jan 2026

Pipeline

Pitch upload

VC/CTO/Product agents

Adversarial CoT

Consensus vote

Strategic report

Pitch Deck Debater

Simulated investor boardrooms with VC, CTO, and Product Lead personas, an adversarial Chain-of-Thought feedback loop, and consensus-voting logic that resolves inter-agent hallucinations.

ImpactBefore → After

1→3: Adversarial investor personas
No voting→Consensus: Voting resolves hallucinations

Pipeline

Pitch upload

VC/CTO/Product agents

Adversarial CoT

Consensus vote

Strategic report

View project

Persona RAGSep – Nov 2025

Pipeline

ChronosGuru

Persona-driven historical RAG platform hitting 92% factual accuracy, holding persona fidelity across 20+ turns, with hybrid keyword + semantic search cutting retrieval latency under 200ms.

ImpactBefore → After

72%→92%: Factual accuracy
~1.2s→<200ms: Retrieval latency
3 turns→20+: Turns of persona fidelity

Pipeline

Document ingest

Hybrid search

Vector DB

Persona prompt

Murf TTS

View project

On-Device Small Language Model Chatbot project visual

Edge AIMar – Jul 2024

Pipeline

On-Device Small Language Model Chatbot

Fully offline SLM chatbot running on-device with no cloud dependency — quantized inference tuned for constrained memory and latency budgets.

0: Cloud calls — fully offline
Quantized: Runs on constrained memory

Pipeline

Model distillation

TFLite conversion

Quantization

Pruning

Local state

View project

Venus Forex Pvt LtdApr – May 2026

Pipeline

FinSight

Real-time forex intelligence platform fusing live rate APIs with an NLP news aggregation pipeline, stateless zero-PII architecture, and programmatic RBI FEMA rate disclaimers.

Zero-PII: Stateless architecture
150+: Currency pairs normalized
12: Active currency corridors

Pipeline

Rate APIs

150+ pairs

NLP news

Zero-PII

React UI

View project

Winner — StratX DevHacks 2025Jan 2025

Pipeline

Restaurant post

Geolocation search

Real-time availability

Verified onboarding

Fresh Start

Award-winning food redistribution platform where restaurants post surplus donations and the community finds them through real-time geolocation search with live availability.

1st: Place — StratX DevHacks 2025
Live: Geolocation availability search

Pipeline

Restaurant post

Geolocation search

Real-time availability

Verified onboarding

View project

Generative AISep 2023 – Mar 2024

Pipeline

AI Video Generation & Analysis Pipeline

Text-to-video pipeline with Whisper transcription, NLP summarization and key-insight extraction, and GPT-driven content generation on a Flask backend.

4-stage: Transcribe → summarize → generate
Whisper: Automated transcription

Pipeline

Text prompt

GPT scripting

Whisper

NLP summarize

Flask pipeline

Pipeline

Halo Trace — AI-Powered LinkedIn Profile Finder

FastAPI + React app with multi-layer text recognition (Gemini Vision with EasyOCR fallback) that extracts names from images and resolves verified LinkedIn profile URLs in seconds.

2-layer: OCR with automatic fallback
Seconds: Image to verified profile

Pipeline

Image upload

Gemini Vision OCR

EasyOCR fallback

Identity search

LinkedIn resolve

View project

OrchestrationApr – Jul 2024

Pipeline

GCP + OpenAI Automation

Multi-API orchestration platform pairing GCP services with OpenAI — predictive weather, multi-participant calendar scheduling, and Gmail automation over OAuth 2.0.

3: Google services orchestrated
OAuth 2.0: Secure delegated access

Pipeline

Weather API

Calendar API

Gmail OAuth

OpenAI

Python orchestration

Pipeline

Neural Style Transfer

Combined content and style features from CNN layers using Gram matrix texture representation and a composite loss function — foundational generative vision work.

Composite: Content + style loss
CNN: Layer-wise feature extraction

Pipeline

Content image

CNN features

Gram matrix style

Composite loss

Optimization

Pipeline

Stock Prediction IoT

IoT-fed real-time market monitoring with predictive price models, multi-source feature engineering, and a Streamlit analytics dashboard.

Real-time: IoT market data feed
Multi-source: Feature engineering

Pipeline

IoT feeds

Feature engineering

Model training

Streamlit dashboard

Pipeline

Air Canvas — AI Virtual Painter

Air-drawing interface built on skin segmentation, contour analysis, fingertip localization, and state tracking for draw, color-select, and erase tools.

3: Gesture-driven tools
Real-time: Fingertip tracking

Pipeline

Webcam feed

Skin segmentation

Contour analysis

Fingertip localize

Virtual canvas

View project

Aura Alarm — AI-Powered Smart Wake-Up Platform project visual

Applied MLSep – Nov 2025

Pipeline

Sleep tracking

Pattern viz

Morning briefing APIs

Web Audio adaptive alarm

Aura Alarm — AI-Powered Smart Wake-Up Platform

Sleep-tracking analytics with pattern visualization, personalized morning briefings from weather/news/calendar APIs, and a Web Audio adaptive alarm that learns from feedback.

3: Briefing APIs blended
Adaptive: Learns from user feedback

Pipeline

Sleep tracking

Pattern viz

Morning briefing APIs

Web Audio adaptive alarm

View project

03 — Experience

The road so far

Enterprise AI at PwC, aerospace-grade delivery at Honeywell, and hands-on production work at applied-AI startups.

Jul 2026 – Present

qBotica, Inc.

Phoenix, AZ

AI Engineer (Part-Time)

Jul 2026 – Present

Built a LiveKit voice agent (web + phone) with HubSpot CRM caller verification that automates end-to-end freight order booking into Degama dtmsWeb via a three-tier pricing engine.
Eliminated hallucinated-price checkout errors on BayOps AI's multi-vendor ordering platform with price-sanity validation plus Twilio/Playwright/Claude Computer Use cart automation and xlwings billing sync.
Improved chatbot response accuracy 30% over semantic and fixed-size chunking baselines with an agentic-chunking RAG pipeline and retrieval tuning.

Jun 2026 – Present

DriverAI

Peoria, AZ · Remote

AI/ML Developer (Part-Time)

Jun 2026 – Present

Achieved 80%+ model accuracy training computer-vision models on live camera-feed data, processing 200,000+ images through an AWS S3 ingestion and training pipeline.
Selected YOLOv12n for production after benchmarking four YOLO architectures on mAP, GPU memory footprint, and inference latency — optimizing for edge constraints over marginal accuracy.

Jan 2026 – May 2026

Honeywell Aerospace

Tempe, AZ

AI/ML Engineer — Capstone

Jan 2026 – May 2026

Delivered a live Mesa Fire Department pilot of FirstWatch, an AI-powered wearable mental-health platform (smartwatch, mobile app, command dashboard), leading a 5-person team through architecture and deployment.
Projected to cut burnout 30% and accelerate critical intervention 40% through a real-time biometric pipeline with predictive stress and crisis-detection models.

Jan 2023 – Jun 2024

PwC

Bangalore, India

Generative AI Associate Consultant

Jan 2023 – Jun 2024

Cut contract-analysis runtime 70% for enterprise legal workflows with production AI microservices on FastAPI and Azure Document Intelligence for clause extraction.
Accelerated semantic search and record retrieval 60% by engineering HyreGPT, an enterprise RAG/LLM pipeline on LangChain, FAISS, and Azure OpenAI.
Reduced inference latency 40% and manual-review overhead 48% via PyTorch quantization and pipeline optimization, while preserving data-privacy compliance.

Jul 2022 – Dec 2022

Atsuya Technologies

Chennai, India

ML Engineer Intern

Jul 2022 – Dec 2022

Achieved 100% uptime with zero cloud dependency by building a real-time YOLO object-detection system on Android (TensorFlow Lite) for Indian Oil Corporation's industrial safety compliance, backed by an offline-first Kotlin inference framework.

04 — Capabilities

The stack I work in

Hover any logo to see it come alive, and switch tabs to see how I actually group my skills.

Python

PyTorch

TensorFlow

LangChain

OpenAI

Anthropic

Hugging Face

FastAPI

Docker

AWS

Azure

Google Cloud

Databricks

PostgreSQL

React

Node.js

OpenCV

Playwright

Streamlit

GitHub

Multi-Agent Orchestration (LangGraph, LangChain)Model Context Protocol (MCP)RAGModel EvaluationVoice AgentsFine-Tuning (LoRA / QLoRA)Transformer ArchitecturesLLMs & SLMsVision-Language ModelsPrompt EngineeringPlaywrightComputer-Use Automation

05 — Recognition

Honors & awards

Recognition for applied AI, secure product thinking, innovation, and professional impact.

Honorable Mention — MLH Innovation Hacks 2.0

April 2026

Arizona State University

Received an Honorable Mention for Voyager AI at MLH Innovation Hacks 2.0.

Best Security Award — Opportunity Hack 2025

October 2025

Arizona State University

Recognized for a secure, impact-focused nonprofit website redesign.

Hackathon Winner — PI Academy

December 2024

Principled Innovation Academy

Winner of the Fall 2024 Problem Solving Event.

PwC Excellence Recognition

2023

PwC

Recognized by leadership for GROQ findings and presentation.

06 — Education

Where I learned the fundamentals

Graduate data science at Arizona State, built on an electronics and computer engineering foundation.

Arizona State University, Tempe

M.S. Data Science, Analytics and Engineering

Aug 2024 – May 2026 · Tempe, AZ

Coursework in statistical machine learning, data mining, deep learning, and large-scale data processing.
Honeywell Aerospace capstone: led a 5-person team building FirstWatch, an AI wearable mental-health platform piloted with Mesa Fire Department.

Vellore Institute of Technology, Chennai

B.Tech Electronics and Computer Engineering

Jul 2019 – Jul 2023 · Chennai, India

Electronics and computer engineering foundation: embedded systems, signal processing, and computer vision.
Published IEEE-format research on multimodal emotion-aware music recommendation.

07 — Profile

Certifications & Research

Certifications

Azure AI Engineer Associate (AI-102)
Advanced MLOps — Databricks
MCPs: Advanced Topics — Anthropic

Publication

Music Recommendation based on emotion detection using facial expressions and speech recognition

M. Roshan Aditya & Sravanakumar Sathish

IEEE-format research paper · 2023

A multimodal recommendation framework that combines real-time facial-expression and speech analysis with AI-generated music suggestions tailored to the listener’s emotional state.

CNNOpenCVSpeech RecognitionOpenAI APISpotify data

Download PDF

08 — Recommendations

What it’s like to work with me

Full recommendations from the managers and colleagues I worked with on applied AI and machine-learning projects.

“I have had the pleasure of working closely with Sravan over the past year and can confidently say that they are a true expert in AI/ML and Generative AI. Sravan possesses a rare combination of technical prowess, creative thinking, and a deep understanding of machine learning algorithms. His ability to conceptualize and implement cutting-edge generative AI models exceeded our project goals, while his collaborative and communicative approach made him an invaluable team member. His positive attitude, willingness to take on challenges, and ability to work seamlessly with cross-functional teams make him an asset to any organization.”

Kokila Natchiyappan

Manager

PricewaterhouseCoopers LLP

Directly managed Sravanakumar

“I had the opportunity to work with Sravan on a project where I focused on research and documentation, and he handled the machine learning engineering. I was impressed by his deep understanding of ML and how he could simplify complex problems into clear solutions. His technical skills are top-notch, and he always delivered high-quality work. Sravan is a great communicator who kept the whole team aligned, and I would recommend him to anyone looking for a skilled and reliable ML engineer.”

Manin Thomas

Associate

PricewaterhouseCoopers LLP

Worked with Sravanakumar on the same team

“I had the distinct pleasure of working closely with Sravan for 1.5 years at PwC India on several Machine Learning and Generative AI proofs of concept. His exceptional technical skills and natural ability to translate business ideas into AI/ML solutions were truly impressive. A skilled orator, Sravan generously shared his knowledge, valued others’ ideas, and maintained a high standard of quality. Any team would be incredibly fortunate to have Sravan, and I wholeheartedly recommend him to any organization seeking top-tier talent.”

JayaSai Surya

Associate

PricewaterhouseCoopers LLP

Worked with Sravanakumar on the same team

Read them on LinkedIn

09 — Contact

Open · May 2026 start

Emailsathishsravanakumar@gmail.com Phone+1 (480) 743-1232 LinkedInin/sravanakumarsathish GitHubsathishsravanakumar ResumeOpen in browser ResumeDownload PDF

Usually replies within 24 hours · AI/ML conversations welcome