AI Research · Data Science · Narrative Analytics
Khansa Khanam
Umar Sultan
I work where research rigor meets real-world clarity: shaping AI pipelines, decoding high-noise datasets, and turning complex systems into stories people can act on.
“What if the real skill is learning how to read noise, uncertainty, and human systems all at once?”
Where data meets research
Computer Science graduate with an IT foundation and business intelligence experience, now working across University at Buffalo labs where AI, society, and scientific modeling overlap. I like building systems that are technically sound and deeply legible.
Research Identity
Full-stack thinking for the full data lifecycle
From unstructured Reddit corpora to cancer cell simulations and executive reporting, I build workflows that move from raw signals to interpretable outcomes. The throughline is always the same: make complexity useful.
AI Research
NLP
Bayesian Optimization
Analytics Storytelling
Working across misinformation, cancer biophysics, and AI & Society.
Experience translating analysis into decisions, not just reports.
Outside The Lab
Writing, nature, and ideas that linger
I write on Medium about AI and society, and I reset through hiking, reading, and long stretches of curiosity-driven thinking.
Working Style
Analytical, visual, and narrative at once
I enjoy the last mile of research: the part where evidence becomes something another human can actually understand and trust.
Communication
Research Ops
Insight Design
Surfacing structure from noisy, high-variance data.
Grounding AI work in experiments, metrics, and reproducibility.
Translating technical results into clearer decisions and stories.
Selected work
A few projects that show how I approach difficult questions: through experimentation, modeling, domain context, and clear communication.
Computational Oncology
Bayesian Optimization for Cancer Cell Migration
Calibrated CompuCell3D simulations across two phenotypes and 32 parameter sets with repeated trials, cutting invalid proposals by 35% while matching experimental invasion behavior more reliably.
View project
Python
Bayesian Optimization
CompuCell3D
Mental Health + NLP
AI4MH Sentiment and Topic Discovery
Fine-tuned RoBERTa on labeled mental-health discussions to reach 87% accuracy across four sentiment dimensions, then extended the analysis to 20,000+ posts with BERTopic for governance-oriented interpretation.
View project
RoBERTa
BERTopic
Reddit Data
Population Health
Physiological Age and Survival Analysis
Built XGBoost and Random Survival Forest models on NHANES data using 20+ biomarkers, reaching a C-index of 0.74 to connect biological aging patterns with Phase 1 enzyme activity.
View project
XGBoost
SHAP
Survival Analysis
My tech stack
The tools I use most often across analytics, data engineering, machine learning, and research strategy work.
Data Analysis
Analysis and visualization
SQL with joins, window functions, and CTEs; Python with Pandas, NumPy, Matplotlib, and SciPy; plus R, Excel, Power BI, Tableau, Google Analytics, and Google Sheets.
SQL
Python
Power BI
Tableau
Data Engineering
Databases and pipeline design
PostgreSQL, MySQL, NoSQL, BigQuery, dbt, Apache Spark, data pipeline design, data cleaning, and reliable data management across messy source systems.
PostgreSQL
BigQuery
dbt
Spark
Cloud
Platforms and integrations
AWS, GCP, REST API integration, and presentation workflows that help move analysis from technical execution into collaborative decision-making.
AWS
GCP
REST APIs
PowerPoint
Machine Learning
Modeling, explainability, and NLP
Scikit-learn, XGBoost, SHAP, PyTorch, TensorFlow, Hugging Face, NLP workflows, and optimization techniques for applied machine learning research.
Scikit-learn
PyTorch
TensorFlow
Hugging Face
Business Strategy
Technical work shaped by business context
Market research across quantitative and qualitative methods, forecasting, competitive intelligence, go-to-market planning, and stakeholder management all shape how I scope and communicate technical work.
Forecasting
Market Research
Go-to-Market
Stakeholder Management
I like connecting models, pipelines, and dashboards to the business questions they need to answer.
The goal is not just insight generation, but helping teams make better decisions with it.
My education
Training across computer science and information technology gave me both engineering discipline and an instinct for making systems practical.
2024-2025
SUNY University at Buffalo
MS in Computer Science, with work spanning machine learning, applied AI research, and data-intensive experimentation.
2017-2021
Anna University
B.Tech in Information Technology, building the technical base that later expanded into analytics, BI, and AI research.
My journey
A timeline of research, analytics, and technical growth that shaped how I work today.
Applied AI Researcher
University at Buffalo, Department of AI & Society. Building reproducible data workflows, analysis artifacts, and research support systems around bias, misinformation, and responsible AI questions.
ML / NLP Research
UB Media Effects Lab. Analyzed 20,000 Reddit posts, fine-tuned RoBERTa models, and surfaced narrative structures with BERTopic for AI governance work.
Biomedical Data Science
UB Systems Biomedicine Lab. Calibrated cancer migration simulations and improved biological model search with Bayesian methods and repeated validation.
Business Intelligence Analyst
SafetyConnect IOTRL. Built forecasting dashboards and strategic analyses that supported business planning during a $1.2M Seed A raise.
A little life
A quieter corner of the portfolio for the parts of me that live outside code, research, and dashboards: hiking, landscapes, and the moments that help me reset.
Let's connect
Open to research collaborations, data science roles, and thoughtful conversations around AI, NLP, social impact, and responsible technology.
khansakh@protonmail.com
The best place to reach me for collaboration, research, or opportunity-driven conversations.
LinkedIn · GitHub · Medium
I share technical work, essays, and ongoing ideas across platforms and long-form writing.