Machine Learning Los Angeles

Saksham Gupta

*****.*********@*****.*** 317-***-**** Github Los Angeles, CA, USA embedding.one WORK EXPERIENCE

Machine Learning Engineer, Deloitte (Los Angeles, CA) January 2024 – Present

- Working with the Audit and Assurance group to help accelerate audit processes using generative AI applications. Implementing long term memory for chat agents, along with more efficient function calling and tool usage.

- Designing efficient deployment workflows for building LLM powered applications focusing on structured data extraction.

- Benchmarking LLM tools like Llamaindex, Langchain, etc. and vector DBs like Chroma, Milvus, Pinecone, etc. for production use cases. AI Governance Machine Learning Engineer, nFactor Technologies September 2023 – December 2023

- As a founding ML engineer, I led the development of an AI/ML Governance platform that enables customers to use the latest LLM services/applications with the assurance of enterprise grade compliance and safety policies.

- Built and scaled RAG (Retrieval Augmented Generation) pipelines, along with spearheading AI safety research.

- Optimizing LLM inference and deployment using techniques like model pruning, quantization, and CUDA optimizations.

- Responsible for end to end software development life cycle (SDLC) – Built CI/CD pipelines using Azure DevOps and Github Actions. Pharma Manufacturing Data Science Intern, Biogen (Durham, NC) June 2023 – August 2023

- Built and deployed an end-to-end pipeline for detecting contradictory statements (NLI) in clinical trial protocol documents using BERT text embeddings and Decoder based large language models (LLMs). Data Science Associate Consultant, ZS Associates (Bangalore, India) August 2020 – August 2022

- Led a cross-functional team to launch a Plotly-Dash-based NLP web application for large scale topic modeling using transformer model embeddings, streamlining organization-wide data visualization, and integrating AWS based CI/CD pipelines and APIs for data access.

- Led efforts in propensity modeling, market segmentation, and linear optimization for leading bio-pharma clients using electronic medical records and insurance claims data; Used explainable algorithms like SHAP, LIME to automate model output explanations from models like XGBoost, Random Forests; Transitioned legacy R workflows to PySpark, resulting in 150% speed improvement, and expanded project scope to cater to three varied drug portfolios.

- Designed a nationwide Transportation optimization platform using agent-based simulation for a premier Japanese pharma client, creating an interactive python backed Tableau Dashboard for transport scenario simulation, using Geo-spatial ML algorithms for optimal treatment center identification.

- Co-hosted a corporate podcast highlighting the career experiences of leading data scientists at ZS and investigated the potential of quantum computing techniques in accelerating core life sciences tasks, as part of an internal initiative. Technology Intern, American Express (Gurgaon, India) May 2018 – July 2018

- Deep Learning-based financial forecasting, integrating time-series ensemble methodologies to enhance forecasting accuracy. EDUCATION

Indiana University - Purdue University, Indianapolis December 2023 Master’s degree (MS), Applied Data Science GPA: 3.96/4

- Coursework: NLP, Biostatistics, Computer Vision, Quantum Computing, Remote Sensing, Database design, Cloud Computing etc.

- Research Assistant:

o Fine-tuning LLMs (Large Language Models) as part of an NLP pipeline to extract adverse drug event data from clinical trial notes. o Evaluating and improving gender/ethnicity based bias detection and mitigation algorithms to make LLMs safer. o Designed and 3D-printed lab containers using Autodesk generative AI and constraint optimization tools – Decreased material wastage by 30%

- Teaching Assistant: 2D Animation, Advanced 2D Animation, Motion Graphics Vellore Institute of Technology September 2020

Bachelor’s degree (BS), Computer Science and Engineering GPA: 3.8/4 SKILLS

- ML sub-areas: Natural Language Processing, Computer Vision, Supervised/Unsupervised learning approaches

- Programming: Python, R, Julia, PySpark, Git, LATEX, MATLAB, MarkDown, HTML, Javascript, Typescript

- Tools: Postgres, SQL, RESTful web services, Neo4j, Hadoop, MongoDB, Elasticsearch, Docker, Kubernetes, FastAPI, Django, Flask

- ML Engineering: MLFlow, Sagemaker, Kedro, BentoML, Dask, Airflow, Langchain, Llamaindex

- ML Packages: Numpy, Pandas, Scikit-learn, Tensorflow, Pytorch, JAX, Huggingface, Scipy

- Visualization: Tableau, Plotly, Matplotlib, Seaborn, D3.js, Power BI, Alteryx, Streamlit, Gradio, Chainlit

- Other: ArcGIS, Google Earth Engine, Github, Microsoft Office

- Communication: English (native), Hindi (native), German (intermediate), Russian (beginner) RESEARCH EXPERIENCE / PROJECTS

- Working with Lakera, a Zurich based AI firm, and the creators of the Gandalf challenge as an AI safety evangelist.

- Participant in the Critical Code Studies Working Group (CCSWG’24), an online think tank for the exploration of code/digital culture.

- Participated in the International Workshop on Pattern Analysis and Applications, Indian Statistical Institute, Kolkata.

- Paper on Explainable ML accepted at the Frontiers of Intelligent Computing: Theory and Applications conference, NIT Surathkal.

- Improved SAR (Synthetic-Aperture Radar) Image de-noising performance by ~15% to remove salt and pepper noise from satellite based remote sensing systems for the Indian Space Research Organization (ISRO).

Contact this candidate