Venkatesh Sanaboina

About

Telecom (Verizon) AI/ML practitioner who built a production multimodal system that ingests messy customer issue reports (calls, chats, emails, screenshots, videos) and turns them into confidence-scored incident summaries with reproducible steps and evidence links. Also built KPI/alarm-to-ticket correlation to rank likely root-cause domains (RAN/Core/Transport), cutting triage from hours to minutes and improving MTTR.

Experience

Education

Key Strengths

• Built and deployed a production AI test case generator converting unstructured customer reports into executable validation workflows
• Designed multimodal incident understanding pipeline (ASR + OCR + evidence linking) with confidence-gated summarization in near real time
• Reduced incident triage time from ~1 hour/hours to minutes by correlating network KPIs/alarms with ticket patterns and ranking root-cause domains
• Strong production orchestration experience using Apache Airflow for end-to-end ML/data workflows (ingestion, validation, Spark/Databricks transforms, training/scoring, publishing to Snowflake/S3)
• Reliability-focused AI evaluation approach: component/workflow/prompt tests, strict schemas, scenario replays, A/B rollouts, monitoring and rollback
• Methodical model/retrieval selection using task constraints and measurable retrieval/answer metrics (recall@k, gold-set accuracy)
• Effective collaboration with non-technical stakeholders to define usable outputs (one-page incident summary, reproducible steps, confidence level, evidence links) and drive adoption