Anandesi Akshay Kumar

Senior AI Engineer with 8+ years of software engineering experience and 4+ years specializing in Generative AI and AI platform architecture. Combines strong backend and distributed-systems expertise with hands-on experience in RAG, LLM evaluation, agentic workflows, and enterprise AI platforms, architecting solutions such as ValOps and ValTest for AI validation and monitoring at scale. Inventor on a pending U.S. patent for real-time GenAI monitoring, with demonstrated impact including an 83% reduction in model-validation cycle time, 50% reduction in AI application development lifecycle, and leadership of a 4-engineer team.

Senior Quantitative Analytics Specialist at Wells Fargo (2023-06 – Present)

• AI Validation Ecosystem: Designed and developed ValOps, a core library that abstracts diverse model frameworks into a standardized interface for enterprise-wide validation of 100+ production models.
• Multi-Framework Integration: Standardized the validation lifecycle across ML, SAS, LLM, RAG, and agentic architectures built on Scikit-learn, H2O, DataRobot, SAS, Hugging Face, and Tachyon.
• High-Performance Data Bridge: Engineered a secure Transponder protocol using Apache Arrow and gRPC, facilitating the compliant, high-speed transfer of large-scale models and datasets across restricted enterprise environments.
• Decoupled Validation Architecture: Developed a specialized Transponder Client that allows independent Validators to connect seamlessly to wrapped models, eliminating direct data dependencies and resolving complex environment fragmentation.
• Standardized Evaluation (ValTest): Built ValTest, a unified library for LLM and model validation that incorporates an LLM-as-a-judge approach to automate hallucination detection, toxicity, and fairness metrics.
• Model Risk Oversight Engine: Led the architecture and development of a real-time Model Risk Oversight Engine for evaluating and continuously monitoring RAG applications and agentic systems, integrating ValOps and ValTest with automated alerting and batch evaluation of custom data.
• Technical Leadership: Led and mentored a team of 4 engineers, guiding the development of scalable validation infrastructure and governance-compliant AI systems.
• Efficiency: Reduced enterprise model-validation cycle time from ~3 months to 2 weeks — an 83% reduction.
• Scalability: Enabled scalable, automated annual validation of 100+ complex models across multiple business lines.
• Architecture: Resolved critical challenges involving environment fragmentation and restricted data access for cross-environment model validation.

Senior Software Engineer – AI & ML at Pegasystems (2021-08 – 2023-06)

• LLM Integration & Efficiency: Led the early adoption and integration of ChatGPT/LLM capabilities into enterprise workflows, enabling conversational AI features for end-users.
• AI Framework Development: Developed internal NLP frameworks within the Pega platform, allowing enterprise users to build and deploy their own ML-powered applications.
• Security & Governance: Designed and implemented a robust PII data masking/unmasking system, ensuring the secure and compliant handling of sensitive information in AI-driven interactions.
• Scalable Infrastructure: Built and deployed scalable microservices using Java Micronaut on AWS Kubernetes, leading to measurable improvements in platform performance and availability.
• Enterprise Logic: Developed reusable components using PRPC, including complex business logic layers, decision rules, SLAs, and automated flows.
• Performance Optimization: Performed deep-level debugging and optimization using PRPC tools such as Tracer, Clipboard, and Rules Inspector.
• Velocity: Cut custom AI development lifecycle by 50% via strategic LLM integration and reusable components.
• Security & Compliance: Secured enterprise conversational AI by automating PII detection and masking.
• Platform Enablement: Enabled users to self-serve and deploy ML capabilities via proprietary NLP frameworks.

Data Science Consultant at Ernst & Young (EY) (2019-12 – 2021-08)

• Platform Architecture: Led the development of an enterprise-level Data Governance and Email Compliance platform integrating data-processing, analytics, and privacy capabilities, reducing operational costs by 20%.
• NLP & Deep Learning: Built and deployed advanced NLP models for sentiment analysis and topic extraction via high-performance REST APIs.
• Petabyte-Scale Pipelines: Engineered highly scalable data pipelines capable of scanning petabytes of unstructured data, achieving a 98.9% accuracy rate in sensitive-data detection.
• Business Growth: Enabled 4x revenue potential by delivering high-accuracy forensic analytics and data-privacy solutions for global enterprise clients.
• Insights & Analytics Dashboards: Developed distributed microservices and interactive dashboards using Elasticsearch and Kibana to enable large-scale data discovery and investigation.

Software Developer at Amdocs India Development Centre (2018-08 – 2019-12)

• Telecom OSS Development: Developed mission-critical telecom OSS applications using Java, Spring Boot, and REST APIs.
• System Stability: Improved system performance and reliability through comprehensive JUnit and Mockito-based testing for third-party integrations.
• Production Optimization: Debugged and optimized production-grade systems to ensure scalability and high availability.

Master of Technology in Data Science – Birla Institute of Technology and Science (BITS), Pilani (2022)

Bachelor of Technology in Computer Science & Engineering – Vardhaman College of Engineering (2016)