AVAILABLE FOR NEW ROLES · 2026 HYDERABAD / REMOTE

Building real-time
computer vision
for defense systems.

Current Focus

ML Engineer at DeepEdge. Leading anti-UAV tracking systems deployed at 3 defense installations. Real-time inference on NVIDIA Jetson edge devices. Sub-pixel object detection < 10ms latency.

Open Source

Contributor to PyTorch Lightning (27K★), Hugging Face (135K★), KerasCV / KerasNLP. Augmentation layers used by 35K+ engineers. 1M+ monthly developer reach.

4.2×
Inference Speedup / TensorRT
5km
Anti-UAV Detection Range
$2M+
Defense Contract Renewal
1M+
OSS Monthly Reach
§ 01 — FLAGSHIP

The work that moves everything else.

DeepEdge · 2025 — Present · Production

Anti-UAV tracking system for defense installations.

  • 01 Pioneered sub-pixel tracking ensemble — Lucas-Kanade + TAPNet + CoTracker — extending detection range from 1.5km to 5km. Directly contributed to $2M+ contract renewal.
  • 02 Architected hybrid CV pipeline (YOLOv8 + Kalman + MOG2/SVM) achieving 95%+ accuracy with 80% reduction in false positives. Outperformed two competing vendors in head-to-head evaluation.
  • 03 Optimized inference 4.2× (250ms → 60ms) through TensorRT quantization, CUDA kernel fusion, and batching — enabled real-time 30fps on NVIDIA Jetson edge devices.
  • 04 Developed 3D world-frame localization integrating 2D detection, monocular depth estimation, camera intrinsic calibration, and IMU sensor fusion. <0.5m localization error validated against ground truth.
  • 05 Established ML CI/CD pipeline reducing deployment time from 2 weeks to 2 days. Mentored 2 junior engineers on computer vision best practices.
1.5km5km

Detection range extension via sub-pixel multi-algorithm ensemble.

250ms60ms

Per-frame inference on NVIDIA Jetson. Real-time 30fps edge.

§ 02 — EXPERIENCE

Paper to production.

MAR 2025 — PRESENT

Machine Learning Engineer

DeepEdge·Defense AI Startup·Hyderabad

Leading end-to-end development of production anti-UAV tracking systems for defense installations. Real-time computer vision on edge. Full ownership from research to deployment.

  • Sub-pixel tracking ensemble: 1.5km → 5km detection range, $2M+ contract impact
  • Hybrid CV pipeline with 95%+ accuracy, 80% false-positive reduction
  • TensorRT + CUDA optimization: 4.2× inference speedup on Jetson
  • 3D world-frame localization: <0.5m error via monocular depth + IMU fusion
JUL 2023 — MAR 2025

Machine Learning Engineer

Deloitte·AI Center of Excellence·Hyderabad

Built production ML systems for enterprise clients — recommendation engines, predictive models, and multilingual RAG pipelines. Shipped measurable business impact at scale.

  • Hybrid recommendation engine (collaborative + content + neural CF) — measurable engagement and revenue uplift
  • Purchase prediction model (XGBoost, 200+ features) — 90% precision, +28% conversion
  • Multilingual RAG pipeline — FAISS vector DB, FastAPI endpoints, Docker deployment
FEB 2023 — MAY 2023

Machine Learning Engineer — Contract

Unify.ai·ML Infrastructure Startup·Remote

Built cross-framework ML infrastructure for the Ivy compiler. Shipped TensorRT converters and rigorous testing for production-grade code used by thousands of engineers.

  • 15+ TensorRT OpConverters: 3× faster inference, 40% memory reduction
  • Hypothesis-based testing suite — 150+ edge-case bugs caught, 95% coverage across 200+ ops
  • Mentored 8 OSS contributors through complex cross-framework codebase
PYTORCH / TENSORFLOW / JAX / YOLOV8 / TENSORRT / CUDA / OPENCV / KALMAN FILTERS / TAPNET / COTRACKER / HUGGING FACE / LANGCHAIN / LANGGRAPH / PEFT / XGBOOST / PYTHON / C++ / DOCKER / KUBERNETES / FASTAPI / JETSON / 
Languages

Python, C++, TypeScript, JavaScript, SQL, Bash.

Deep Learning

PyTorch, PyTorch Lightning, TensorFlow, JAX, Scikit-learn, XGBoost, LightGBM.

Computer Vision

YOLO (v5/v7/v8), Detectron2, OpenCV, Kalman filters, TAPNet, CoTracker, thermal imaging, sub-pixel tracking.

Edge & Optimization

TensorRT, CUDA kernel fusion, quantization, pruning, NVIDIA Jetson, <10ms real-time inference.

LLMs & GenAI

Hugging Face Transformers, PEFT, LangChain, LangGraph, RAG, RLHF (PPO / DPO), vector databases.

Systems

Docker, Kubernetes, FastAPI, CI/CD for ML, AWS, PostgreSQL, Redis, FAISS.

§ 03 — SELECTED WORK

Things I've built in the open.

LLM RLHF Tuning — PPO & DPO

From-scratch RLHF pipelines using PPO and DPO for LLM alignment. Comparative analysis of reward-model-based PPO vs Direct Preference Optimization — trade-offs in stability, compute efficiency, and output quality.

PyTorch TRL Transformers RLHF

Custom Vector Database

Lightweight vector DB built from scratch using Hierarchical Navigable Small World (HNSW) graphs. No LangChain, no Pinecone — just the algorithm and NumPy. Minimal dependencies, easy integration.

HNSW NumPy ANN Search

Paper Implementations in DL

From-scratch implementations of foundational deep learning papers in PyTorch. Not API wrappers — actual algorithm reimplementations to understand the math underneath.

PyTorch Research DL

Flight Scanner — RAG Analytics

Natural language analytics over structured flight data. RAG architecture with DeepSeek-R1-Distill-LLaMA-70B and LLaMA 3.2:3B. ChromaDB retrieval, FastAPI serving.

RAG DeepSeek-R1 LangChain FastAPI

Amazon Product Classification

Fine-tuned ALBERT on 500K Amazon listings across 27 categories with 1:150 class imbalance. 89% F1 — beat BERT by 12% and production baseline by 10%. TensorRT deployment.

ALBERT PyTorch TensorRT

ChildMind Anxiety Prediction

Gated RNNs analyzing children's physical activity patterns for early detection of problematic internet use. Healthcare ML with real clinical relevance.

TensorFlow RNN Healthcare
§ 04 — OPEN SOURCE

Contributions to foundational ML infrastructure.

Project
Stars
Contribution
Link
PyTorch Lightning
★ 27K+
Contributions to training framework used by 50K+ practitioners.
PRs ↗
Hugging Face
★ 135K+
Contributions across the Transformers ecosystem.
PRs ↗
KerasCV / KerasNLP
★ —
Augmentation layers used by 35K+ engineers in production.
PRs ↗
Unify.ai / Ivy
★ —
15+ TensorRT OpConverters; 3× faster inference, 40% less memory.
PRs ↗
§ 05 — RECOGNITION

Competitions & fellowships.

10/100
Challang Global Hackathon

Top 10 teams worldwide. Built disaster response system using satellite imagery + EfficientNet — 91% IoU.

18/2,500
HackerEarth ML Challenge

Top 1% ranking among 2,500 participants in a machine learning competition.

12/1,000+
Cipla Datathon

Top 1.2% in pharma/healthcare data science competition.

500/global
LinkedIn LIFT Scholar '22

Top 500 women technologists globally. 0.5% acceptance rate.

650+
LeetCode — Top 5%

Consistent problem-solving practice. Data structures, algorithms, systems.

GHC '23
Grace Hopper OSD Mentor

Mentored 15+ contributors on OpenSSF project. 25+ merged PRs during the event.

§ 06 — PUBLICATIONS

Undergraduate research.

2020

Privacy-Preserving Cloud Computing

IEMATICS 2020 · Homomorphic Encryption
Read ↗
2020

Machine Learning in Medical Diagnosis

SPECTRUM 2020 · Applied ML for Clinical Diagnostics
Read ↗
§ 07 — CONTACT

Have an interesting problem?
Let's talk.

Location
Hyderabad, IN