AI / ML · Case Study · Led by Varun Vashisht (PMP · CBAP · PMI-PBA)
AI/ML Resource-Allocation Proof of Concept
Engineered a Python-based proof of concept that learned project-resource fit signals, accelerating programme staffing decisions across the portfolio.
The challenge
Programme managers were spending days manually matching projects to available resources. A time-consuming process that created bottlenecks at the start of every engagement. The portfolio had grown to a scale where intuition-based staffing was no longer viable.
The approach
Designed and built a Python-based ML proof of concept that learned project-resource fit signals from historical programme data. Integrated Azure OpenAI to enrich project requirement descriptions before scoring.
- Extracted and structured historical staffing data from PMO records.
- Trained a similarity model using scikit-learn on project attributes and resource skill profiles.
- Built a lightweight interface for programme managers to get ranked resource recommendations instantly.
- Validated PoC results with PMO stakeholders. 80% reduction in time-to-match vs. manual baseline.
Outcomes
- 80% Faster project-to-resource matching vs. manual process
- PoC Validated with PMO leadership for broader rollout
- Azure OpenAI Integrated for requirement enrichment and scoring
Stack & methods
- Python
- Machine Learning
- Azure OpenAI
- PMO Ops
- Pandas / scikit-learn
Role on this programme
Varun Vashisht owned this programme end-to-end as the AI / ML delivery lead. Covered scope definition, stakeholder alignment, vendor governance where applicable, backlog and RAID management, and executive reporting against the 80% faster project-to-resource matching outcomes above. The work draws on the same 14+ years of enterprise programme delivery that underwrites his current Senior Program Manager engagement.
Relevant credentials: PMP (PMI), CBAP (IIBA), PMI-PBA, TCS Certified Generative AI Practitioner. Tech & method stack used here: Python, Machine Learning, Azure OpenAI, PMO Ops, Pandas / scikit-learn.
Recruiter or hiring manager? Let's talk.
If you're staffing a similar programme (ai / ml transformation, enterprise rollout, or AI-enabled delivery), the fastest path is a quick intro.