Available for opportunities  ·  Bengaluru, India

Rohan
Mulay

B.Tech CSE student at PES University building production-grade AI systems — real-time computer vision pipelines, LLM-powered backends, scalable REST APIs, and edge-deployed ML models that work in the wild.

View My Work Resume rohanm1307@gmail.com
3+ Internships
2 Years Experience
10+ CV Models Shipped
5+ Languages
Scroll to explore

I'm an AI/ML Engineer & Full-Stack Developer who builds things end-to-end — from training deep learning models to shipping them behind production-grade FastAPI backends, containerized with Docker, running on edge hardware or in the cloud.

My interests span Computer Vision, LLM & GenAI integrations, ADAS systems, and real-time inference on constrained devices. I care about the full pipeline: data, architecture, optimization, deployment, and observability.

When I'm not training models, I explore biomedical AI, systems programming, and building developer tools — because great engineers understand the whole stack.

Education
B.Tech CSE  ·  PES University  ·  2022–2026
Current Role
AI Engineer Intern @ Cautio (Acq. BYTES)
Focus Areas
AI/ML  ·  LLM & GenAI  ·  Computer Vision  ·  Edge AI  ·  ADAS
Location
Bengaluru, Karnataka, India
Contact
rohanm1307@gmail.com  ·  +91 7483994847

Work Experience

AI Engineer Intern
Cautio  Acq. BYTES
Bengaluru, Karnataka
Oct 2025 – Present
  • ADAS Safety Suite for Dashcams: Engineered a full ADAS safety feature stack for commercial dashcam platforms implementing Forward Collision Warning (FCW), Lane Departure Warning (LDW), Driver Monitoring System (DMS), and Headway Monitoring System (HMS) via real-time multi-camera CV pipelines.
  • Integrated IMU sensor fusion with camera feeds for robust, weather-resilient ADAS performance; leveraged temporal smoothing and Kalman filtering to suppress false positive alert rates.
  • Pan-India Driver Face Recognition: Architected and deployed production-grade facial recognition using ArcFace / FaceNet embeddings with FAISS vector indexing and anti-spoofing liveness detection, rolled out across thousands of commercial vehicle operators pan-India.
  • Optimized inference with TensorRT INT8 quantization on NVIDIA Jetson Orin NX, achieving sub-50 ms end-to-end latency.
  • Refined the 2-wheeler ADAS perception stack with blind spot monitoring and object localization via homography transformations; built lightweight YOLOv11 pipelines with optical flow–driven motion analysis.
  • SalesOps Automation: Designed ML-driven automation including lead scoring models, automated pipeline analytics, and CRM data enrichment, reducing manual effort and improving conversion tracking.
ADAS ArcFace FaceNet FAISS TensorRT Jetson Orin NX YOLOv11 Sensor Fusion FCW / LDW / DMS Liveness Detection Kalman Filter OpenCV
ADAS & Machine Learning Intern
Moonrider.ai
Bengaluru, Karnataka
Apr 2025 – Oct 2025
  • Contributed to an ADAS system for autonomous electric tractors, focusing on robust off-road perception in unstructured environments.
  • Developed a semantic segmentation pipeline for lane and drivable area detection using data augmentation to handle diverse terrain conditions.
  • Optimized the perception pipeline for embedded systems, ensuring reliability in real-time operations under compute constraints.
  • Built ensemble models and LSTM-based architectures for accurate SoC/SoH prediction and anomaly detection from multivariate time-series battery data.
Semantic Segmentation LSTM Autonomous Vehicles SoC/SoH Prediction PyTorch Embedded Systems Real-time Inference

Selected Work

Featured Projects

01 · Biomedical AI

Integrative Multi-Omic Biomarker Discovery for NAFLD

Multi-omic ML framework using Graph Neural Networks and co-attention mechanisms to predict NAFLD progression from MRI scans and multi-omic profiles. Leveraged SHAP and LIME for XAI-driven interpretability, enabling personalized risk stratification and biomarker discovery.

GNNs TensorFlow SHAP LIME MONAI Python
02 · Systems & Networking

Secure Cloud Storage via Socket Programming

Architected a secure, multi-threaded client-server system supporting remote file upload, download, delete, and listing over TLS-encrypted sockets. Implemented concurrent multi-client handling with threading for high scalability on Linux VMs.

Python OpenSSL / TLS Linux Multithreading Networking
03 · Computer Architecture

Configurable Multi-Level Cache Simulator

Fully configurable multi-level cache simulator supporting Direct Mapped, Fully Associative, and Set Associative caching policies. Integrated LRU, FIFO, and Random replacement strategies with accurate hit/miss rate simulation using real-world memory access traces.

Python Linux Cache Architecture LRU / FIFO / Random

Technologies

My Tech Stack

Toolkit

Technical Skills

Languages
Python C/C++ Java SQL JavaScript HTML/CSS
Deep Learning & CV
PyTorch TensorFlow TensorRT ONNX YOLO OpenCV Scikit-learn MONAI
LLM & GenAI
LangChain OpenAI API Hugging Face RAG Pipelines Vector DBs Prompt Engineering
Backend & APIs
FastAPI REST APIs Node.js React PostgreSQL Selenium
ADAS & Perception
Sensor Fusion Kalman Filter Optical Flow Depth Estimation Homography FCW / LDW / DMS
Edge AI & MLOps
NVIDIA Jetson CUDA TensorRT INT8 Docker MLflow Git Linux

Let's Build
Something Great

Open to AI/ML engineering roles, research collaborations, and interesting problems at the intersection of computer vision and real-world deployment.