~/about

About

avatar

Mohit Appari

Data Scientist
APD State of Florida
mailgithublinkedin

I'm a Data Scientist with 4+ years of experience building deep learning, forecasting, and statistical systems — neural networks, LLM-powered applications, and time series models — to solve real problems in finance, government, and research. I currently work at the Agency for Persons with Disabilities (State of Florida), building spending and budget forecasts and financial analysis pipelines over high-volume transactional data. Before that, I wrote production software at S&P Global, researched equivariant neural networks for materials science at Florida State University, and along the way developed a deep interest in prediction and forecasting. I like sitting at the intersection of deep learning, math, and finance, and I'm always chasing problems that are hard enough to be interesting.

I'm eager to contribute to AI engineering initiatives that bridge deep learning research with real-world impact.

Day to day, I work across the ML stack — training neural networks, building LLM-powered applications, and developing statistical and time series models (PyTorch, scikit-learn, LangChain, R, Python) that turn messy, high-volume data into accurate predictions and forecasts.

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
‣
Process and analyze 200K+ records across Medicaid, vendor, and client systems serving 60,000+ individuals — merging data from SQL Server, S3, and internal systems into unified datasets, and leading development of Power BI dashboards and automated reports (Python, SPSS) for executive leadership and agency staff.
‣
Optimize SQL queries and data pipelines through indexing, partitioning, and caching strategies, reducing report execution time by 83% (60 minutes to under 10 minutes).
‣
Conduct statistical analysis using SPSS, SAS, and R to uncover billing anomalies, forecast financial trends, and deliver compliance insights, improving audit accuracy and budgeting. Lead ad hoc and recurring data support for cross-functional teams, cutting manual workload by 40% through reusable code pipelines.
~~~
AI/ML Researcher @ Florida State University
June 2024 – September 2024
‣
Researched open-source generative AI models and algorithms (GANs, VAEs, RNNs) to explore synthetic data generation for material property prediction and battery optimization applications.
‣
Implemented and benchmarked NequIP (MIT open-source) for molecular stability prediction across an 11,000+ material dataset, improving prediction consistency by ~15% during initial validation.
‣
Collaborated with PhD researchers in Chemistry to align data science workflows with experimental research goals, enhancing interdisciplinary research on AI-driven material discovery and energy storage.
~~~
Software Developer @ S&P Global
January 2022 – August 2023
‣
Identified a migration opportunity as an intern — auditing S&P's legacy .NET document-processing workflows against open-source alternatives (regex, pdfplumber, Camelot, Tesseract, Textract) — then led a team of 5 through the migration, owning the architecture and rollout of a Python-based extraction platform across multiple business domains, cutting licensing costs 80% and improving extraction speed 30%.
‣
Built the pipeline layer on Docker/AWS to process 10,000+ financial and automotive documents daily with retry logic, structured logging, and fault-tolerant processing, reducing manual extraction time by over 90%. Combined OCR with CNN-based classification to turn unstructured PDFs and scanned documents into structured data, improving entity recognition accuracy on financial documents.
‣
Engineered the platform for long-term maintainability with modular, reusable components, CDC-based incremental updates, and 80% test coverage, backed by GitHub Actions CI/CD for reliable releases.
~~~
Software Developer Intern @ LG Electronics
May 2021 – August 2021
‣
Built a real-time drowsiness detection system using Python, OpenCV, YOLO, and Raspberry Pi, achieving 90% facial landmark accuracy to improve driver safety in logistics operations.
‣
Designed a CNN-powered logistics tracking system integrated with REST APIs and PostgreSQL, reducing package sorting time by 25% across the logistics workflow.
‣
Awarded 3rd place in LG's internal Ideathon competition for the logistics tracking system built during the internship.
~~~
Software Developer Intern @ ITC Limited
January 2021 – March 2021
‣
Developed a master data management system using Python (Flask, REST API) and MySQL with multi-role access control across admins, users, and vendors.
‣
Implemented role-based authentication and approval workflows to improve data security and streamline user operations across business units.
‣
Automated product tracking and data update processes, reducing manual workload by 30% and enhancing data accuracy across multiple business units.
~~~

Education

Masters in Data Science @ Florida State University
Aug 2023 – May 2025
‣
Cumulative GPA: 4/4
‣
Coursework: Statistics, Mathematics, Research, LLMs, Analytics, AI, Databases, Big Data, Business and Finance
~~~
Bachelors in Computer Science @ Christ University
Aug 2018 – May 2022
‣
Cumulative GPA: 3.85/4
‣
Coursework: Data Structures, Operating System, Software Architecture, Programming, Networking, Communication, SDLC
~~~