Tanisha Pradeep Bhola

About

Accenture engineer with hands-on experience taking an NLP sentiment analysis system from prototype to production, emphasizing robustness to noisy data, scalability, and observability (dashboards for latency/error/throughput). Also supports customer-facing teams with demos and PoCs, translating client requirements into secure, scalable architectures and troubleshooting LLM/agent workflows via logs and step-level traces.

Experience

Education

Key Strengths

• Takes NLP/LLM prototypes to production with strong engineering hygiene (APIs, refactoring, monitoring)
• Improves robustness to noisy real-world data via stronger preprocessing and validation checks
• Performance and scalability optimization (stateless services, background processing, horizontal scaling)
• Strong observability mindset (logging, dashboards, key metrics like latency/error rate/throughput)
• Real-time debugging of agentic/LLM workflows using logs, traces, and step-level analysis
• Effective technical communication through demos/workshops tailored to mixed developer and stakeholder audiences
• Bridges technical and customer-facing teams to address scalability/integration/security concerns and drive adoption
• Built and maintained high-traffic FastAPI microservice on AWS with JWT auth and API Gateway
• Reliability engineering: strict timeouts, exponential backoff retries, centralized exception handling, structured HTTP errors
• Serverless delivery on AWS (Lambda/API Gateway/DynamoDB) with CloudFormation IaC and CI/CD automation
• Cost/scaling optimization in serverless: stateless functions, cold-start/package-size tuning, right-sized memory/timeouts, CloudWatch monitoring, API Gateway usage plans
• ETL ownership with AWS Glue: schema cataloging via crawlers, schema evolution via versioning/backward-compatible changes, in-job data quality checks
• Operational incident ownership for batch pipelines: CloudWatch alerting, log/metric triage, partition-level reprocessing, improved validation/limits/timeouts
• SQL performance tuning with measured impact (PostgreSQL query reduced from ~20s to <3s using EXPLAIN ANALYZE, composite indexes, and query rewrites)
• Legacy modernization: monolith-to-microservices migration with parallel runs, regression tests, feature flags, and stakeholder sign-offs