~/about

About

avatar

Mohit Appari

Data Scientist | Quant Research & Systematic Trading
APD, Florida
Say Hi!

I'm a Data Scientist researching systematic trading strategies, grounded in academic factor literature (Jegadeesh-Titman momentum, Fama-French/Carhart factors, Hamilton regime-switching models) and validated through rigorous backtesting. I built a Hidden Markov Model-based market regime detector that outperformed static strategies on a walk-forward out-of-sample test (Sharpe 0.769 vs. 0.641, drawdown -25.6% vs. -33.7%), and a full-stack quantitative research platform combining a multi-factor scoring model with a backtesting engine, regime filter, and live paper-trading execution across 500+ tickers. Looking for exciting opportunities to work and collaborate with bright minds and explore the world of data together.

If you're embarking on an exciting project or seeking fresh perspectives, don't hesitate to reach out via Mail or Whatsapp.  I'm always eager to connect, exchange ideas, and explore new avenues of exploration and growth together.

Check out some of my photography work on VSCO.



About this site

Welcome to my home on the internet. This site functions as a blog/portfolio, a place to share code and thoughts. Opinions of my own.

I learnt how to build this site from the most awesome people in the community:

Experience

Data Scientist  @  APD
June 2025 – Present
‣
Performed forecasting and budgeting analysis on ~$200M in monthly financial transactions, producing charts, supporting data, and reports used for rate-change decisions and financial planning. Conducted research and trend analysis across a 60,000+ client base (state waiver program), tracking client movement between services, service utilization patterns, and data integrity issues.
‣
Built and maintained 40+ automated data pipelines processing 300K+ records (Python, SQL Server, Azure Data Warehouse, Docker), ensuring reliability for downstream forecasting and financial analysis.
‣
Optimized SQL-based analytical workloads through indexing, partitioning, and caching, reducing query latency by 83% (60 min → under 10 min), enabling faster iteration on forecasting models.
~~~
Research Assistant  @  Florida State University
June 2024 – September 2024
‣
Researched generative models (GANs, VAEs, RNNs, Diffusion) under Dr. Bin Ouyang for synthetic molecular data generation, focused on discovering stable, sustainable materials for lithium-ion battery applications using AI-driven compound synthesis.
‣
Benchmarked and fine-tuned deep learning models (NequIP) via transfer learning across an 11,000+ material dataset, improving prediction consistency by 15% through iterative experimentation and hyperparameter optimization.
‣
Designed and executed controlled experiments across multiple model architectures using A/B testing and statistical comparison methods, tracked via MLflow, ensuring rigorous, reproducible model evaluation and informing experimentation design practices.
~~~
Software Developer  @  S&P Global
January 2022 – June 2023
‣
Built a production-grade data extraction platform processing 10,000+ documents/day, combining OCR and CNN-based classification models to convert unstructured PDFs and scanned documents into structured datasets for downstream analysis and modeling.
‣
Engineered cloud-based data pipelines (AWS S3, Athena, BigQuery) to ingest, transform, and serve large-scale datasets, enabling analysts and downstream models to access up-to-date structured data within minutes.
‣
Designed a modular, reusable Python extraction framework with CDC-based updates and 80% test coverage, and developed REST API integrations with schema validation and CI/CD (GitHub Actions) to improve data quality and deployment reliability.
~~~
Software Developer Intern  @  LG Electronics
May 2021 – August 2021
‣
Built a proof-of-concept LG WebOS logistics platform in a 2-week ideathon, integrating real-time shipment tracking, live delay and congestion feeds, and a computer vision drowsiness detection system (Python, OpenCV, YOLO) achieving 90% facial landmark accuracy for driver safety.
‣
Designed and integrated a CNN-based object recognition module with REST APIs and PostgreSQL for automated package sorting, reducing sorting time by 25% across the logistics workflow.
‣
Awarded 3rd place out of all competing teams. Presented the solution to LG leadership and engineering teams, receiving commendation for practical application and technical innovation.
~~~
Software Developer Intern  @  ITC Limited
January 2021 – March 2021
‣
Built a Master Data Management system using the MERN stack with role-based authentication (admins, users, vendors) to manage commodity tracking and application workflows across multiple ITC business units.
‣
Implemented approval workflows and role-based access controls to enhance data security and streamline user operations across business units.
‣
Automated product tracking and data updates, reducing manual workload and improving cross-unit data accuracy through API integrations and backend synchronization.
~~~