Relevant Experience

"You miss 100 percent of the shots you never take"

Leading a team of 10 hyper active genai developers
Design and develop applications to analyse root cause of system instability with GenAI,
Reduced MTTR by 15% for infra systems instability with root cause analysis
Fine-tune Large Language Models to provide recommendation for specific tasks
Benchmark performance of LLMs on different hardware devices
Train SLMs for RCA
Deployment of SLMs on on-prem environment

Working on hyper-personalisation domain
Using Genetic Algorithm to dynamically action text and call-action-button on images
Improved user retention for stakeholders by 4.5% sending personalised notifications using reinforcement learning
Generation of unique notification image payload to increase user engagement with the help of genetic algorithms
Faster payload generation by 10% via improved database optimized queries
Exploration on user recency and RFM

• Extraction of high value analytics from Legal Corpus

• Building robust ML pipelines and providing concurrent API end points

• Improving speed of text-based captcha solving by using SOTA object detection

• Improvising fuzzy text matching to be 25% faster on large scale text matching on quoted text from two legal cases

• Building case summarization tool of relative chronology of case proceedings using SOTA transformers based models

• Building a first person reasoning model to identify the polarity of speaker

• Extraction of common provisions, clauses, and data points from contracts

• Building, scaling and deploying an ensemble model of transformer and supervised machine learning algorithms to extract

information in form of one-class classification

• Improving in-house API to serve multiple requests and made it faster by 10%

• Built a sentence similarity model to find similar text wrt. context of the text using SOTA NLU techniques

Developing Deep Learning models using LSTM, GRU and xgboost to predict the rise and fall of NIFTY.
Working on high frequency time series data to automate the data gathering and prediction.
Developing barriers for uncertainty quantification to external noises.
Tech Stack - Python, Deep Learning, Bayesian Techniques

Financial data modeling.
Statistical analysis on long term prediction of hedge-funds like S&P 500 with 92% accuracy.
Testing and developing models for stock price prediction.
Tech Stack - Python, Machine Learning

Process economic/finance domain algorithms and methods for feature extraction.
Use LSTM, Dynamic Time Warping, Clustering and scalable methods to build prediction system.
Create a GDP prediction system with 5% more accurate than IMF.
Modeling uncertainty quantification using Bayesian methods to give the uncertain quantity for one phase of prediction.
Tech Stack - Python, Machine Learning, Deep Learning, DTW