Avinash Anand

av.iitk@gmail.com

+91-7420-929-270
Senior AI Engineer
linkedin.com/in/avinash-anand-46836b92

github.com/AVIN8233

WORK EXPERIENCE SKILLS

Senior AI Engineer, Private Equity Performance Machine Learning Deep Learning Generative AI
Improvement
Alvarez and Marsal Transformers LLMs Python Pytorch MLOps
Achievements/Tasks ( Nov'2023 - Present )
HuggingFace Chatgpt PySpark Statistics
Lead the development of Diligence GPT, a finance large language
model, incorporating AFR & unstructured for precise text
extraction and implementing enhanced chunking logic.
Executed advanced hybrid search methodologies, including
similarity search, semantic reranking, and hybrid search, PERSONAL PROJECTS
culminating in an impressive 85% accuracy on diverse document Effect of Quarterly Results on Stock Prices (05/2023 - 10/2023)
formats such as PDFs and PPTs. Conducted a comprehensive analysis to evaluate the influence of quarterly
Conducted comprehensive parameter grid exploration to fine- financial results on stock prices and formulated alpha strategies aimed at
outperforming the NIFTY index by 4%.
tune chunk size and the number of chunks (k) for optimal model
Data Collection: Gathered and organized financial data, including quarterly
performance & search accuracy. reports, earnings statements, and historical stock price data
Orchestrated the seamless deployment of the product on Hypothesis Testing: Formulated and tested hypotheses related to how
Microsoft Azure, establishing a robust model monitoring pipeline various financial metrics (e.g., QoQ Profit, earnings per share) influence stock
within the MLOps framework for continuous optimization and price fluctuations. Fitted XGBoost model to conclude high impact attributes
performance enhancement.
Multi label Video Classification using Deep Learning
Currently driving the evolution of chunking strategies through the (08/2016 - 12/2016)
adoption of more comprehensive algo like Apache Tika and PDF Classification of Movie Trailers into multiple genres(21 different genres) using
Layout, alongside innovative approaches such as knowledge Deep Nets( LSTM ) in Keras, Hamming loss for MutiLabel Classification
graph utilization to elevate the capabilities of LLMs
Hierarchical Reinforcement Learning Using Deep Q-Learning
(01/2017 - 04/2017)
Data Scientist To make the player learn in a goal-directed environment with fixed
Microsoft, Hyderabad reinforcement-reward updates.
Achievements/Tasks ( Oct'2019 - May'2023 ) To maximize the total reward and minimize (to an extent) the number of steps
in achieving the final goal over the period of time (i.e., the number of games).
Field Issues Classification Model: Constructed a resilient NLP
model for precise MXPA Field Issues category prediction,
harnessing BERT embeddings derived from the raw inquiries,
complemented by a clustering algorithm to discern the optimal EDUCATION
no. of topics/ clusters - A semi supervised task.
BackLog Forecasting Project: Led the initiative to precisely
MS-BS Dual Degree, Mathematics and Scientific
forecast project burn hours, encompassing at project and Computing
portfolio levels using tailored clustering solution. IIT Kanpur
Introduced & developed Dynamic interval estimates model to 07/2012 - 06/2017, 7.6/10.0
solve Random Order of Magnitude(ROM) price of a project
empowering Sales to strategically target based on advance MBA Minor, Industrial Management and
revenue estimates. Engineering
Microsoft Azure, Office, Dynamics – Next Best Solution(NBS) IIT Kanpur
recommendation: Recommend Services solution to Microsoft
05/2015 - 06/2017, 8.7/10.0
partners to drive Azure Consumption based on prior purchase
pattern, dollars spent & Markov Transition Probability Matrix.
Led MLOps implementation, orchestrated Data and Model
Governance, and managed data drift for diverse regression, ACHIEVEMENTS
classification, and time series models, streamlining the ML
Submission for Microsoft Machine Learning and Data Science
lifecycle and fostering a culture of automation and optimization.
Conference Accepted (04/2021)
Our Paper on Next-Best Consulting Solutions Recommender got accepted as Talk in
Data Scientist Microsoft Data Science Conference
Sears Holdings India, Pune One MXPA Award, Microsoft (05/2022)
Achievements/Tasks ( July'2017 - Oct'2019 ) For playing key role in Intelligent MLOps – Automated ML Model Governance and
Development of deep NLP model based on embeddings and Business Value Quantification Capability
LSTM to generate actionable insights from Member feedbacks
Saving $300K yearly by building inhouse NLP Model. PjM as a Service Program Awards, Microsoft
For solving Project management problem and equipping PjMs with Data Intelligence
Developed an advanced recommender system leveraging
Autoencoders to optimize SKU recommendations across diverse Secured AIR 1960 in IIT-JEE’12, and AIR 224 in GATE
product categories. Platform: GCP -Google Cloud Platform MATHEMATICS 2016