Saketh Bollina

• Partnered with legal operations and finance leadership to map manual contract review workflows onto an LLM-based automation architecture; fine-tuned BERT, GPT-4, and LLaMA with QLoRA for contract classification, obligation extraction, and executive summarisation, improving accuracy by 28% and cutting analyst review time from multiple hours to under 30 minutes across thousands of monthly contracts. • Architected production RAG pipelines using LangChain and vector databases (FAISS, Pinecone) grounded in the organisation's verified policy library, legal precedents, and approved contract templates; embedded RAGAS retrieval evaluation to give compliance and risk teams an auditable quality signal before acting on any AI output — cut hallucination rate by 40% and materially reduced legal review overhead on AI-generated recommendations. • Built multi-agent LangGraph orchestration with conditional routing, risk-tiered escalation, and human-in-the-loop approval gates replicating the decision logic of senior reviewers — eliminated 60% of manual processing touchpoints across intake, triage, and sign-off stages, directly converting analyst capacity from document triage to strategic contract negotiation and commercial risk analysis. • Quantified and presented platform ROI to senior leadership: automated workflows cut per-document processing cost by ~35%, reduced turnaround from multiple business days to under four hours, and enabled 3× throughput without proportional headcount growth — accelerating revenue-generating contract execution and improving working capital cycles. • Deployed MLOps infrastructure on AWS SageMaker with MLflow, DVC, and Kubernetes auto-scaling sustaining 99.9% uptime at 10,000+ concurrent requests; implemented SHAP/LIME explainability and drift monitoring reducing silent model failures by 45% and producing audit-ready governance documentation — decision logs, bias reports, model cards — for regulated enterprise deployment.

• Led discovery workshops with C-suite and VP-level operations leaders at Fortune 1000 clients to translate vague strategic mandates into KPI-aligned ML problem definitions with agreed success metrics before any model was trained — drove 22% average accuracy improvement and 30% reduction in manual operational costs per engagement by solving the right business problem from the outset. • Deployed a fine-tuned BERT multi-label sentiment and intent classification system for a financial services client processing thousands of daily support interactions; surfaced dissatisfaction signals and escalation risk 3× faster than weekly legacy reporting — enabled retention outreach within 24 hours of a negative signal and measurably reduced high-value account churn across targeted customer segments. • Architected a RAG-based editorial research system (LangChain, OpenAI API, Pinecone) for a media client managing years of archived content; applied TruLens retrieval quality validation before launch — reduced research time per article by 55% and directly increased content production throughput without additional editorial headcount. • Delivered a collaborative filtering and LLM-augmented recommendation engine for a retail client; validated via A/B test against a holdout control group — 25% improvement in click-through rate and 18% reduction in customer churn, producing directly attributable incremental revenue and a quantified business case for full-scale commercial rollout. • Translated all model outputs into board-level Power BI and Tableau dashboards mapping predictions directly to client P&L drivers — churn risk tiers, forecast variance, pricing elasticity, campaign ROI lift — improving AI recommendation adoption by 40% and directly influencing client decisions on pricing strategy, inventory allocation, and marketing investment

• Acquired comprehensive expertise in finance functions, including working capital finance, transfer payments, and import-export procedures. • Correlated internal ratio analysis with a financial analyst firm's report to understand the company's financial health. • Reviewed funding mechanisms, sanction letters, and margin calculations to understand the company's credit facilities and financial strategies. • Analyzed documentation, payment terms, and foreign exchange transactions for importing green coffee beans and other materials.

• Assisted the marketing team in tracking basic campaign metrics (engagement, leads, conversions) and compiling weekly performance reports for internal review. • Used Google Analytics dashboards to monitor website traffic and summarize key insights for the team. • Helped organize customer and lead data in spreadsheets/CRM systems to support follow-ups and improve data accuracy. • Coordinated with the finance team to reconcile marketing expenses and prepare simple spend summaries. • Supported preparation of presentation slides summarizing campaign performance and marketing KPIs for management meetings. • Gained exposure to how marketing data, customer tracking, and financial reporting connect to business decisions.

Awarded $18,000 Director’s Achievement Scholarship