Programming Languages:

Technologies:

Cloud Platform:

Blockchain:

Core Distributed Systems Engineer-Versatus Labs (August 2022-Current)

• Experienced in L1 blockchain development, focusing on computationally lightweight block mining and consensus algorithms. 
•  Significant contributions to the implementation of DKG, Threshold signatures, Quorum Election, and Token Emission Methodology. -
• Designed and developed Left Right Concurrency over Mempool, enabling efficient broadcast over UDP through the use of RaptorQ in the networking stack. -
• Led the creation of a Decentralized Task Scheduler, optimizing network utilization with the FarmerHarvester Quorum. 
•  Currently engaged in research and exploration of innovative techniques to detect and address malicious network activity.
  
• Worked on metrics integration using Prometheus,also worked on storage layer for IPFS .
• Some of my contributions: https://versatus.io/library/farmer-harvester-bft.pdf

Rust Engineer -SupraOracles ( July 2021 -July 2022 )

• Collaborated with world-renowned PhDs from some of the top universities to develop a Hybrid Consensus (Proof of DTS) for Layer 1 blockchain at a fast-growing startup. 
•  Designed and implemented modules for critical components, including nested DKG, Transaction Engine, Tribe Clan Model, Batch Proposer, Block Proposer, Threshold Signer, and BLS Sig Aggregator. 
• Actively participated in building and refining the Hybrid Consensus algorithm for robustness and scalability. - Played a key role in optimizing the performance of the consensus mechanism for high throughput and low latency.
• Conducted thorough testing and debugging to ensure the reliability and security of implemented modules.
• Contributed to architectural discussions and provided technical insights to enhance the overall blockchain ecosystem.
• Actively participated in code reviews and mentored junior team members to maintain code quality and best practices.

SuperMoney (Technical Architect)  June 2016 -June 2021

• Worked as Technical Architect for Fintech services development.
• Built a powerful backend infrastructure utilizing MicroServices with REST/gRPC APIs.
• Developed a Real-Time Notification Engine to send customer reminder notifications for repayments and eligibility for new loan products. 
• Implemented a system with a calibrated ML model that reduced the likelihood of loan defaults by 40%.
•  Created a Deep Learning model for document identification (e.g., PAN card, cheque) and performed OCR on PAN, Aadhar, Cheque, Insurance documents, etc.
•  Constructed a text analytical engine that profiles expenses and cash inflow from SMS messages.

Software Engineer-BookMyShow (Aug 2014 - May 2016) 

•  Solved problem of Cold-Start Recommendation Engine for recommending music using Audio Data. 
•  Solved problem of Seat Occupancy by pinging 5k+ Cinemas for real-time seat allocation. 
• Worked on solving large-scale data engineering problems by building large pipelines. 
•  Built backend microservices using gRPC/REST in Golang. 
•  Was also part of DevOps for deployment of Machine Learning Services.

BI Consultant Team Computers-Mercedes (Aug 2013 - Feb 2014)

• Worked as a trainee, gaining hands-on experience in Business Intelligence (BI) tools. 
• Focused on QlikView and utilized it to build visually appealing and interactive dashboards for marketing analytics. 
• Explored data insights, enabling the marketing team to make data-driven decisions and optimize their strategies.
• Gained proficiency in data visualization techniques, transforming complex data into intuitive and actionable visuals.
• Developed a solid foundation in BI concepts and tools, paving the way for future projects and career growth.

Empowering Document OCR: Harnessing Python for Accurate Text Extraction

Deep Neural Networks for Accurate Document Detection

CardioDetect: An Innovative Approach to Cardiovascular Disease Detection

Objective: To design a system for detecting cardiovascular disease.

• Led end-to-end deployment, testing, monitoring, and retraining efforts for a document detection project, encompassing every stage from Exploratory Data Analysis (EDA) to Model Creation and Deployment.
•  Assumed full responsibility for the detailed EDA phase, identifying key insights and patterns to inform the subsequent model creation process. 
• Successfully developed and deployed the document detection model, rigorously testing its performance to ensure high accuracy and reliability.
•  Implemented robust monitoring mechanisms to continuously track the model's performance and detect any deviations or anomalies. 
• Achieved a remarkable Recall of 82% using F2 score as the evaluation metric, showcasing the model's effectiveness in correctly identifying relevant documents. 
• The project's GitHub repository (https://github.com/vinay10949/CVD) serves as a testament to the dedication and excellence put into this endeavor, providing valuable insights for the broader community.

DeepGuard: Empowering SmartBackground Verification with Neural Networks

Short-Term Loan Default Prediction for Enhanced Financial Stability

Objective: To develop a robust and predictive model capable of early identification and classification of loan defaulters, empowering financial institutions with actionable insights to mitigate risk, improve decision-making

• Led data engineering and data pipelining efforts, ensuring seamless data processing and analysis for our customer base.
•  Conducted detailed analysis of customer data, extracting valuable insights to inform business strategies and decision-making processes. 
• Successfully trained and tested a predictive model to identify first-time customers at risk of defaulting on a 5000 Rupee product loan. 
• Achieved an exceptional F-Beta Score of 0.79, showcasing the model's high accuracy and reliability in predicting loan defaults. 
• Spearheaded a groundbreaking initiative resulting in a significant 40% drop in loan default rates, enhancing financial stability and customer trust. 

Music Recommendation Engine

Objective: Pioneering User-Centric Music Experience: Develop a Cutting-Edge Solution to Address the Cold Start Challenge, Elevate User Engagement, and Acquire Actionable User Data on the Music Application.

• Successfully tackled the Cold Start problem and improved user engagement on a music application. 
• Utilized Music Information Retrieval (MIR) from the Vienna University library, incorporating features like rhythm patterns and temporal descriptors (approx. 2000 dimensions per song). 
• Implemented Locality Sensitive Hashing and the Epsilon Greedy Approach for robust music recommendations. 
• Technology stack: Python, Protobuf, Konga, Redis. 
• Achieved real-time music recommendations for a user base of 50 million users, solving the Cold Start challenge effectively.

Masters in Computer Applications 2010-2013 

Computer Science Major (Full Time)

Bachelors in Computer Science 2007-2010

Computer Science Major