Anmol Mittal

Conducted an extensive exploration of LLMs and cutting-edge prompt engineering techniques to generate high-quality summaries from client call transcripts. Orchestrated the launch of a pilot application for 50+ users. Developed an automated evaluation framework and real-time LLM quality control layer. Enhanced existing frameworks by developing an n-gram keyword extraction system and fine-tuned a BERT classifier for client interest extraction from emails.

Developed In-Season Growth Stage and Yield Prediction models for Crop Science. Set up monitoring tasks on Grafana for ML models on Algorithmia. Developed a crawling architecture for financial news and a probabilistic topic model using Pyro LDA. Worked on extracting named entities and causal relationships using roBERTa and transformer-based models.

Developed an end-to-end news crawling architecture using selenium and bs4, storing data in MongoDB. Collaborated with cross-functional teams to maintain data quality for 60L+ news records. Performed sentiment analysis, trending news extraction, and news categorization.

Developed an internal web API for a business dashboard to automate the EDA process. Analyzed 300M+ user base to estimate potential business opportunities using PySpark on Hadoop framework.