R&D AI Perception

Mohanen Balasubramanian

Senior Lead Engineer (AI R&D)

Professional Summary

Innovative AI R&D Lead with 10 years specializing in Computer Vision, Deep Learning, and production-ready ADAS solutions for automotive leaders. Proven contributor to patented technologies, including Multi-Task Origin Agnostic BEV Perception. Expertise in zero-to-production algorithm development, multi-task neural network optimization, and robust perception systems.

Experience

Senior Lead Engineer (AI R&D)

Valeo India Pvt Limited • Chennai, India

FEB 2021 - PRESENT
  • Spearheaded AI R&D for advanced ADAS perception systems delivered to global automotive leaders.
  • Co-invented patented BEV perception models; optimized multi-task neural networks for edge inference constraints.
  • Engineered zero-to-production computer vision algorithms spanning parametric lane detection, targetless trailer tracking, and dynamic camera calibration.
  • Delivered highly optimized, low-resource perception models achieving strict KPI parity and deployment standards.

Project Engineer

E-con Systems / VisAi Labs • Chennai, India

MAY 2016 - JAN 2021
  • Engineered core algorithms targeting stereo depth perception, 3D point cloud reconstruction, and customized camera calibration.
  • Led end-to-end computer vision product development lifecycles from rapid prototyping to production-scale deployment.
  • Mentored engineering personnel; structured technical task planning for cross-functional product execution.

Patents & Inventions

2025

[Title Generalized for IP] Advanced BEV Perception System

2025

[Title Generalized for IP] Long Range Surround View Perception

2025

[Title Generalized for IP] Intelligent 3D Lane Guidance System

2023

Real-Time Extremely Low Resource Environment Based Multi-Task Learning Guided Emergency Vehicle Light Detection

2022

Gooseneck and Fifth Wheel Trailer Hitching and angle prediction

Projects & Innovations

Multi-Task Origin Agnostic BEV Model

VALEO 2026 - Present

Co-invented patented BEV perception model. Achieved single-grid KPI parity. Developing commercial truck multi-ROI solution.

Multi Task BEV Network

VALEO 2025 - 2026

Unified BEV model for parametric lane & object detection.

Lane Detection

VALEO 2024 - 2025

Point-based BEV lane detection via 4 surround-view cameras projected to BEV grid. Built core implementation from scratch.

Multi Task Learning

VALEO 2023 - 2024

Unified network for soiling, segmentation, object detection, trailer tracking, and EVD.

Targetless Trailer Tracking

VALEO 2021 - 2022

Vision+kinematics UKF fusion for ADAS. Custom client architecture.

vMeasure - Object Dimensioning

VIS-AI LABS 2019 - 2020

Stereo-camera 3D dimensioning (L/B/H). Custom min bounding box algorithm.

Stereo Presentation Attack

VIS-AI LABS 2019

Anti-spoofing for ATMs. Prototyped VGG16 model classifying human faces using depth inputs.

TaraXL - Stereo Camera

VIS-AI LABS 2018 - 2019

Developed semi-global matching algorithm. Optimized with CUDA for 10x speedup on Jetson Nano.

EsomGears & Health Monitor

VIS-AI LABS 2016 - 2018

OTA & device monitoring for Linux/Android. Engineered AWS IoT cloud infrastructure via MQTT.