Resume

📄 Download a PDF version of my Resume.

Rishikesh Ajay Ksheersagar

Houston, TX | +1 734 489 2596 | rishiksh@umich.edu LinkedIn: https://www.linkedin.com/in/rishikeshksheersagar/ Website: https://rishiksh20.github.io/

Profile

Skills: Machine Learning (ML), Large Language Models (LLMs), Agentic AI, LangChain, AutoGen, Retrieval Augmented Generation (RAG), Deep Learning, Natural Language Processing (NLP), Anomaly Detection, Regression Analysis, Statistical Inference, Reinforcement Learning (RL), Information Retrieval, Bayesian Inference, Agent Based Models, CI/CD

Languages: Python (Pandas, Dask, NumPy, ScikitLearn, Tensorflow, PyTorch, Keras, NLTK, Spacy, StreamLit), SQL, R, PySpark, SAS, C++

Tools / Platforms: Azure, Snowflake, Hadoop, GCP, AWS, Jenkins, Tableau, PowerBI

Education

University of Michigan — Ann Arbor

Masters in Data Science August 2023 – May 2025 | GPA: 4.0/4.0

Relevant Coursework:

• CSE 595 – Natural Language Processing
• ECE 598 – Foundations of Large Language Models
• CSE 545 – Machine Learning
• SI 650 – Information Retrieval
• STATS 510 – Probability Distributions

Savitribai Phule Pune University

Bachelor of Engineering – Computer Engineering June 2015 – June 2019 | GPA: 3.7/4.0

Professional Experience

LatentView Analytics

Data Scientist Houston, TX, USA | November 2025 – Present

• Developing Generative AI powered summarization and BERT-based case-note classification pipelines to transform large-scale telemetry device-health and reliability data into concise, structured, and actionable reports for engineering and leadership of a Fortune-80 technology devices firm.
• Building scalable quality and performance analytics systems using PySpark to monitor boot time, battery wear, and thermal behavior across millions of personal system devices.
• Designing automated serial-number level alerting and triage frameworks with Power BI dashboards and PowerApps to support week-over-week prioritization and executive decision-making.

Ecological Servants Project

Data Analysis and Research Intern Ann Arbor, MI, USA | August 2025 – November 2025

• Led environmental analytics and digital growth initiatives by assessing GA4/SEO dashboards, conducting keyword cannibalization audits, and developing competitor–sponsor intelligence tools to guide content and outreach strategy.
• Designed scalable data pipelines integrating web analytics, search performance, and scraped datasets to optimize nonprofit visibility, donor engagement, and conservation impact.

University of Michigan

Research Assistant Ann Arbor, MI, USA | May 2024 – May 2025

• Engineered a scalable AutoGen evaluation framework that stress-tests LLMs for deceptive (scheming) behavior across thousands of realistic SWE scenarios, dynamically exercising multi-tool chains (Git, shell, file, and 2k+ live APIs) to measure safe tool use at scale.
• Conducted NLP research on “Climate Change, Demographic Shifts, and Socio-Political Stability in Sub-Saharan Africa” under the Minerva Initiative. Automated metadata retrieval, PDF scraping, and text extraction for 50k+ research papers. Analyzed 20k+ causal sentences linking climate change and social unrest using POS tagging and LLMs.

Graduate Student Instructor

• Conducted weekly lab sessions for 60+ students in the QMSS 301 course during Fall 2024 and Winter 2025 semesters covering Geospatial Analysis in R, Predictive Modeling and Sentiment Analysis in Python, Web Scraping, and Research Methodologies.

Mu Sigma Inc.

Apprentice Leader

• Managed two teams consisting of 8 data scientists working with Fortune-100 clients in telecom and healthcare domains, spearheading the growth and management of engagements generating $1M annually.
• Achieved a 30% decrease in probable outages by designing an unbalanced multi-class classifier using RxMER data, stacking XGBoost and sequential neural network models to precisely identify causation of modem network impairment anomalies in near real-time.
• Developed a capability-building PoC tool to simulate patient journeys in clinical trials by integrating therapeutic area, site, PI, patient, and trial attributes utilizing Bayesian networks and agent-based models enabling proactive planning and mid-trial adjustments for Phase 3 clinical trials.
• Drove RFP connects with CXOs of two Fortune-100 telecom clients showcasing deep domain expertise and strategic solution alignment.

Decision Scientist

• Led a team of 7 data scientists in identifying key features for degraded network service for the data science and data engineering teams of a Fortune-100 telecom client.
• Delivered 98.7% accuracy in detecting degraded network service events by conducting statistical analyses and hypothesis testing across 7 datasets including cable modem registration, speed tests, and modem utilization while designing an anomaly detection framework.
• Reduced execution time by 60% by enabling digital transformation for a legacy store planning tool for the FP&A team of the world’s largest home improvement retailer by migrating from a SAS–FileZilla system to a machine learning backed Python–GCP solution.
• Enabled historically aligned financial plans for 8 retail metrics via constrained optimization and time series modeling.
• Designed a failure tracking system that reduced tool failures by 50% and decreased debugging time by 75% by automating root cause analysis and dynamic failure correction.
• Created 7 Tableau dashboards to provide insights and anomaly detection for financial planning metrics.

BMC Software

Project Intern Pune, India | August 2018 – April 2019

• Implemented a proof-of-concept private blockchain with a voting-based consensus mechanism using Hyperledger Composer alongside a traditional structured database in the backend of a legacy ITSM platform.

Academic Projects

PapeRet (Sept – Dec 2024) Designed a research paper retrieval system processing 98,000+ academic papers using recursive metadata extraction, web scraping, PDF download, and text extraction. Leveraged LLaMA for Retrieval-Augmented Generation (RAG) to generate summaries. Achieved MAP@10 of 0.539 and NDCG@10 of 0.81.

Register Augmented LLM Fine-Tuning (Oct – Dec 2024) Developed a register-augmented fine-tuning approach for LLMs enhancing global context management and interpretability. Implemented RegBERT for QA tasks improving F1 and Exact Match on the TyDiQA GoldP dataset with attention analysis using LRP and Integrated Gradients.

Few-Shot Preference-Based RLHF (Jan – May 2024) Implemented few-shot preference-based reinforcement learning algorithms including MAML, iterated MAML, and REPTILE to optimize human feedback efficiency on MetaWorld datasets achieving ~90% reduction in training time.

Is it Easy to be Multilingual (Nov – Dec 2023) Studied transfer mechanisms in mBERT highlighting syntactic, morphological, and phonological similarities as predictors of cross-lingual transfer and proposed a framework achieving 62.5% accuracy in selecting optimal source language.

Honors and Awards

• Mu Sigma Inc. SPOT Award recipient for three consecutive years (Aug 2022, Aug 2021, Oct 2020) for exceeding project goals and delivering high-impact solutions.