U
Uzma Langde
About
Detail
Pune, Maharashtra, India
Résumé
Jobs
verified_user
0% verified
Senior GenAI - Al Platform EngineerCodvo.aiJul 2023 - Apr 2024 (10 months)Built and maintained 100+ REST APIs for document insights, Al job execution, and multi-tenant access. Designed an Al job orchestration layer supporting execution, retries, versioning, and lifecycle management. Integrated Keycloak-based RBAC to enforce fine-grained access control across tenants and services. Implemented secure document storage and metadata management using Azure Blob Storage. Leveraged event-driven communication to decouple services and scale backend workflows. Orchestrated review workflows, reruns, and job state tracking with telemetry-enabled logging. Built an agent orchestration layer to manage agent lifecycle, task delegation, and state persistence for long-running Al workflows.- Senior GenAI - Al Platform EngineerCodvo.aiDec 2022 - Feb 2026 (3 years 3 months)Designed and built enterprise-scale Agentic Al and Document Intelligence platforms using LLMs, RAG, vector search, and event-driven architectures. Architected and operationalized RAG pipelines with semantic indexing and embedding-based retrieval to enable low-latency, context-aware document search and Q&A. Implemented agentic Al frameworks where specialized agents collaboratively handled retrieval, reasoning, validation, OCR selection, and response synthesis for multi-step workflows. Developed LLM-powered skills (summarization, classification, entity/fact extraction, reconciliation, Text-to-SQL) using OpenAl and Gemini models. Built scalable backend and microservices architectures using FastAPI, async processing, and NATS for distributed Al
- Design And Development EngineerSiemens HealthineersJun 2021 - Nov 2022 (1 year 6 months)I am a data and software developer at Siemens Healthineers, where I have Developed ETL Data Integration pipelines in SAS DI Studio that extract and transform raw healthcare equipment data from Oracle Data Lake, to create tables, Union all views and Data Marts in the Data Warehouse Teradata. Orchestrated the pipelines using Apache Airflow. Migrate and transform (ELT) all Oracle's raw data into Snowflake. Developed ETL Data Quality Automation application using Python and SQL that does the tests like data validation, metadata testing, incremental data load check, etc. for each Data Mart, Table and Union All View. Created dashboards from the resulting reports in Power BI. Use Azure Blob for storage and CI/CD in Azure DevOps for deployment. Pres
Education
verified_user
0% verified
- CPost Graduate Diploma in Big Data AnalyticsCDAC KP, BangaloreJul 2020 - Jun 2021 (1 year)76%
- DMaster's in Computer ScienceDr. BAMU, AurangabadApr 2018 - May 2020 (2 years 2 months)71%
- DBachelor's in Computer ScienceDr. BAMU, AurangabadApr 2015 - May 2018 (3 years 2 months)78%
Projects (professional or personal)
verified_user
0% verified
- EEnterprise GenAI & Document Intelligence Platform (NelO.ai)Feb 2025 - Current (1 year 5 months)Architected an enterprise-grade RAG pipeline for ~10K documents, achieving sub-3s query latency with high contextual relevance. Developed LLM-powered Al skills (summarization, classification, entity/fact extraction, reconciliation) using OpenAl and Gemini. Built a Text-to-SQL system enabling natural-language queries over structured data using semantic parsing and embeddings. Designed semantic indexing and retrieval using Qdrant to support document search and cross-document context linking. Implemented scalable microservices architecture with FastAPI, NATS, and async processing for Al workloads. Added OpenTelemetry-based observability to monitor latency, failures, and Al pipeline health. Implemented an agentic Al framework where retrieval, r
- SSub-Batch Document Processing SystemApr 2024 - Mar 2025 (1 year)Designed a distributed OCR and text extraction pipeline for large-scale unstructured document ingestion. Integrated LLM-based OCR and vision models to improve extraction accuracy across scanned and semi-structured files. Implemented sub-batch processing to balance throughput, fault tolerance, and resource utilization. Automated document-to-Markdown conversion to standardize downstream indexing and summarization. Built event-driven error handling and retries to ensure reliable recovery from partial failures. Enabled asynchronous orchestration for parallel document processing at scale. Introduced agent-driven workflows where agents selected OCR strategies, validated outputs, and triggered reprocessing based on confidence signals. Key Skills: