Akaash Nidhiss Shanmugapandian

• Surfaced ~$1.4B in savings opportunity for a large PE client by redesigning material harmonization end to end: a four-step pipeline (cheap filter, LLM relevancy pass, embeddings clustering, LLM sub-grouping with confidence scores), later generalized into the firm-wide tool.
• Drove ~$234M in benchmarked savings through the firm's internal benchmarking and value-discovery platform: spend benchmarks, a savings library from hundreds of past initiatives across a dozen clients, and a supplier discovery agent; ~$234M realized against a ~$5.7B spend base.
• Opened a ~$3M+ opportunity, incl. ~$500K in purchase management, with a purchase-request review prototype for a ~$100B retailer, a working tool built from a workflow problem in days.
• Built a self-serve procurement workflows platform that turns repeated spend cleanup into one-click pipelines, used across telecom, retail, machinery, logistics, and banking. Handles ~100K supplier-entity harmonizations and billions of dollars of spend categorization a day; on-demand workflows return in 2-4 hours, an 8x increase in project delivery capacity. 50+ deployments, 100+ daily users.
• Built the data asset behind it: refreshable categorization and harmonization pipelines plug into a growing repository of AI-generated, SME-validated categorizations and supplier golden records, so every project run enriches the corpus and the next starts more accurate.
• Built a multi-tenant commodity should-cost platform that splits category spend into cost drivers and tests supplier prices against input-cost movement. A mining deployment on ~$401M of spend with an allocation reference showed ~$19.3M of value on a ~$663M baseline.
• Cut benchmark export from ~1 day to 10-15 minutes by building supplier intelligence and peer benchmarking over ~1.5M suppliers and 200+ companies, replacing brittle text-to-SQL with per-use-case agents over a live data lake.
• Engineered core parts of a pre-engagement rapid assessment tool that helps prospective clients spot immediate opportunities in their own spend data: on a sample run, ~8-11% savings surfaced on ~$16.4M of spend across 131 suppliers, with Word and Excel outputs.
• Defined product vision, ran biweekly demos with Partners and CXOs, authored decision memos and roadmaps, and mentored junior engineers.

• Built an LSTM-based predictive-maintenance model on industrial IoT sensor data, achieving 100% breakdown prediction at ~10% false flags.
• Reduced the working sensor set from 52 to 2 through clustering and correlation analysis.

• Facilitated workshops and developed go-to-market presentations for cybersecurity offerings aimed at company leaders and C-level executives.

CGPA 7.80 · Top 10 Finalist, PayTM PM Case Study Competition

Gold Medalist, Scholarly Proficiency (5 years)