review draft · not a final resume
Mohammad Haider
data science / applied ML evaluation
Statistics-trained engineer working across data systems, observational analysis, and independent model experiments. Nearly four years at Microsoft and UVA building and validating reporting systems and decision-support analytics.
Experience
Independent Software Engineer
- Indigo Circuit: built a tournament analytics pipeline using Python, DuckDB, dbt, Glicko-2, and Flask. Experimental empirical-Bayes deck estimates describe observed matches, not causal card effects. Source
- C-1N: trained and evaluated a six-legged PPO policy in MuJoCo. The mean policy reached 0.888 m/s; 24 fixed flat-ground evaluations had zero falls. These are simulation results. Evaluation record
- LM Lab: built a PyTorch transformer in an AI-assisted learning project and compared five depths from 1 to 8 blocks over three paired seeds at equal update budgets. Mean held-out loss fell from 3.529 to 3.337 nats/token; larger models used more compute. Experiment report
University of Virginia · IT Analyst Intermediate, BI Developer
- Interviewed department leads and delivered and supported 20+ SQL-backed Tableau dashboards; wrote shared metric definitions.
- Reconciled conflicting purchase-order states across financial systems and surfaced the deltas and their financial impact. Contributed to a team effort that improved organization-wide dashboard load times by about 20%.
Microsoft · Software Engineer I
- Owned a zero-downtime SSAS Multidimensional-to-Tabular migration for financial reporting; defined parity checks and kept the legacy service running through cutover.
- Extended automated financial-value reconciliation and monitored Azure data pipelines. Telemetry showed up to 200× faster filtering and approximately 4× faster Excel loading.
Selected data project
Log Pose: built a dated evidence store and source-backed graph over 14 pinned inventories, retaining source hashes and separate capture and arrival times. Co-listings are observations, not verified company relationships. Source
Methods and tools
SQL · Python · R · pandas · NumPy · PyTorch · MuJoCo · dbt · DuckDB · PostgreSQL · Tableau · Azure Data Factory · Synapse · Kusto · SSAS · Git · CI
Education
Virginia Tech · B.S. in Statistics · 2021
Use your browser’s print command to save this review draft as a PDF.