[email protected] | +91 8767102424
GitHub | LinkedIn |
StackOverflow
- Passionate computer vision developer with proven experience of creating and deploying models related to object detection, instance segmentation, face recognition/detection from scratch along with active contribution to open-source projects
Sr. Machine learning Engineer (May 2021 - Present)
- Currently working on building object detection pipelines for various use cases
- Worked on 3D object detection algorithms
- Worked on image segmentation problems
- Implemented anomaly detection algorithms for time series data
Computer Vision Engineer (October 2019 - May 2021)
- Lead the Face Recognition project for Mumbai Metro (MMRDA) Literature survey and training models from scratch with large-scale data along with OpenVINO optimization and hardware setup. Data augmentation to tackle racial bias (specific to Indian faces) and low lighting without sacrificing LFW accuracy (99.3%)
- Face detection model on the lines of LFFD (A Light and Fast Face Detector for Edge Devices) for real-time crowd monitoring.
- Packaged deep learning models (SR, face mask detection, social distance monitoring etc.) into production level C++ SDKs for clients oriented towards deployment on edge devices like routers (Client - Cisco)
- Created Super Resolution models using Residual Dense Block architecture for entertainment industry and mobile phones (lightweight models by pruning and quantization)
- Created benchmarking tools to assess model performance in terms of VMAF, PSNR, SSIM etc. for super resolution applications
Developer Python | Machine Learning (March 2017 - October 2019)
- Developed TV Feed Analyzer an automated commercial extraction and story segregation platform based on broadcast video (news feed) for Times Now and Zee News(Zee Hindi and Wion) by deploying:
- Machine learning techniques (HOG+LSTM, CNNs, PCA, linear regression) for anchor detection, face recognition, scrolling text band detection etc.
- Traditional CV methods (shot detection, Hough Line transform) for scene classification, footage box detection etc.
- Audio analysis (MFCC based Voice Activity Detection)
- Built scalable back-end from scratch for "TV Feed Analyzer" using AWS Lambda, EC2 and Mongo DB (system design, DB architecture, cron automation etc.)
- Deployed end to end machine learning pipelines on AWS
- Found solutions to problems in the video processing domain to optimize workflow and avoid bottle necks related to speed as well as compute costs on AWS (multi-processing in serverless environment)
- Joined as junior Python Developer to produce maintainable and well documented code for analytical and research-oriented problems related to computer vision, audio and video processing
MATLAB Developer (July 2016 - August 2016)
Advanced MATLAB Project implementation dealing with:
- Image Processing
- Machine Readable Zone Detection
- Optical Character Recognition and other important topics related to physics
- Writing tutorials and code (in MATLAB) spanning concepts such as Monte Carlo simulation, Harris corner detection, Radon transform, etc.
Don Bosco Institute of Technology, Mumbai
Bachelor's degree (Electronics and Telecommunications
Engineering)
2012 - 2016
- State Estimation and Localization for Self-Driving Cars
- Practical Machine Learning with Tensorflow 2.0
- Machine Learning for Engineering and Science Applications (NPTEL)
- Programming, Data structures and algorithms using Python (NPTEL)
- Foundations of Wavelets and Multirate Digital Signal Processing (NPTEL)
- Sparse Optimization for Signals & Systems (Indian Institute of Technology, Madras)
Industry Knowledge: Computer Vision / Machine Learning / Deep Learning o OpenVINO optimization (C++ & Python) o AWS (AWSCLI, S3, Lambda, EC2, CloudFront) o Data Structures and Algorithms o Basic front-end skills o Github o 3D Computer Vision o Lidar data processing
Programming: Python (Proficient) o C/C++ (Intermediate) o CUDA programming o JavaScript o PHP o MATLAB o VHDL
Machine learning frameworks: PyTorch (torchvision, torchaudio) o Keras API (Tensorflow backend) o Tensorflow 2.0 o Scikit Learn
Interpersonal Skills: Public Speaking o Pitching products (sales pitch)
- Face recognition using Arcface and Face detection using LFFD
- Image Classifier for fine grained classification and imbalanced dataset
- Voice Activity Detection based on MFCC/spectrogram classification
- Automated audio levelling for byte footages (NAT sound)
- Trending article suggestion based on K-means clustering
- Intelligent seam carving & Background removal
- Blurred text detection using LSTM
- Object detection pipeline (SSD/Faster-RCNN)
- U-Net text detection
- Speaker at DevOps India Summit-21 Topic-“Chaos engineering for Artificial Intelligence using MLOps approach”
- Keyboardist, song writer @ Keyband - Official band of Kalyan Eparchy Youth (Performed with artists including Alphonse Joseph and Rexband)
- Trinity Guildhall certified Keyboardist
- Served as Choirmaster for St. Joseph's Church, MC Road (2018-2022)
- Invited by Don Bosco Institute of Technology, Mumbai to conduct a session on "Machine Learning - Current Trends & Avenues"
- Secured 1st prize in "Instrumental Event" at CREXTAL 2014
- Led the St. Joseph's Church, MC Road Church Choir to victory at Fr. Sunny Memorial Church Choir Competition in the year 2018
- Secured 2nd place in "Debate" at TALENTIA 2015
- Secured 5th place in "One Minute to Fame (Keyboards)" at TALENTIA 2015
- Secured "ELITE" status in NPTEL certification course "Practical Machine Learning with Tensorflow 2.0"