I'm an AI engineer focused on explainable AI: models that can show why they made a prediction. At iLink Digital I build forecasting, knowledge-graph and GenAI systems for enterprise clients across five industries, with an explanation behind every prediction.
I'm also a researcher, with 10 peer-reviewed papers (9 as first author) and 4 under review, including a benchmark for explainability methods, plus work on LLM calibration and low-resource NLP. I studied Computer Science (AI & Data Science) at MIT World Peace University.
Skills
Open full screen →Every skill, linked to the work it was used in. Explainable AI is highlighted. Drag the timeline, filter by domain, or click a skill to see where it was applied.
Open the skills graph →Experience
iLink Digital· AI Innovation and Research Consultant 2026 – Present
- Demand forecasting for a U.S. diagnostics network: LightGBM across 2,292 sites, 770K half-hourly p10/p50/p90 predictions (out-of-fold MAE 0.31), SHAP explanations on every forecast.
- Wait-risk classifier (AUC 0.78) flagged 122 sites early; rebuilt anomaly detection to cut false positives from 193 to 52; built a 29K-scenario what-if engine.
- iWeave ontology accelerator: AWS Neptune connectors, ontology validation, versioning, entity resolution and bi-temporal tracking; ran client workshops.
- Document intelligence for a U.S. steel manufacturer: Azure Document Intelligence into Cosmos DB with GraphRAG retrieval.
iLink Digital· Software Engineer Trainee 2025 – 2026
- Led Microsoft-partnered satellite methane-monitoring pipelines (TROPOMI, PySpark, Delta Lake); calibration cut error from ~20× overestimation to near-parity with airborne benchmarks.
- BI migration of 450+ enterprise reports: lineage analysis cut duplicate semantic models by 23%.
- Co-built Migrate IQ (FastAPI, GPT-4o), cutting manual migration testing by up to 90%. 2nd place, iLink hackathon.
iLink Digital· Consultant Intern, Microsoft Fabric 2025
- Built ETL pipelines, semantic models and Lakehouse dashboards with Microsoft Fabric, PySpark and Power BI across finance, transport and aviation datasets.
Society of Women Engineers· Technical Head → Technical Advisor 2023 – 2025
- Led a 20-member team at MIT-WPU and ran programming and AI seminars.
Freelance· UI/UX Designer 2023 – 2024
- Designed client interfaces from wireframes to live prototypes.
Research
View all →Projects
View all →
Book Ratings Prediction
ExplainableSix leakage-aware models over 342k Goodreads books, explained with SHAP and served in R Shiny.
AniOntology AI
ExplainableAnime knowledge graph with a formal OWL/RDF ontology; every recommendation comes with a "Why?" explanation.
AniRestore
Custom Triton kernel for Real-ESRGAN: 53% less compute on the fused step, a measured 1.6% end-to-end speedup.
MoE Emotion Predictor
Real-time Japanese emotion recognition from text and speech with a learned mixture-of-experts router.
