Meghana Pola

Proficient Python developer with expertise in backend systems and data pipelines. Skilled in FastAPI, Docker, and Kubernetes, driving efficiency and optimizing existing infrastructures.

Software Engineer: Toucan Payments | Hyderabad, India                                                     Apr 2023 - Present 

• Designed and implemented a Data Lake solution integrating MongoDB and remote filesystems, enabling centralized storage and streamlined data access.
• Built and optimized automated data ingestion and movement pipelines using Apache NiFi, Python, and Apache Airflow, improving processing efficiency and workflow automation.
• Implemented Trino as a distributed query engine to enable efficient data retrieval and analytics across stored datasets.
• Developed RESTful APIs using FastAPI to support data ingestion, retrieval, and cross-functional access to Data Lake services.
• Designed and deployed a conversation summarization system with a unified FastAPI endpoint to store raw lead conversations and return AI-generated summaries.
• Integrated MongoDB (Replica Set with TLS authentication) for secure storage and management of structured and unstructured conversation data.
• Implemented structured prompt engineering logic and integrated Ollama with the Qwen 2.5 model for local LLM-based summarization.
• Built a real-time payment orchestration system to route transactions dynamically based on cost, latency, and success rate, reducing transaction failures.
• Developed routing logic and integrated multiple payment gateways to automatically select the optimal gateway per transaction request.
• Created and transformed synthetic datasets to simulate real-world transaction scenarios for business logic validation and testing.
• Deployed applications on a Kubernetes cluster to ensure scalability, high availability, and reliable production performance.

Datalake:

• Spearheaded successful implementation of Datalake solution, integrating MongoDB and remote filesystems.
• Engineered a highly efficient data movement pipeline using Apache NiFi and Python automation, reducing data ingestion time by 40%.
• Implemented Trino query engine, enabling retrieval and analysis of data with improved response time.
• Orchestrated data workflows with Apache Airflow, automating over 100 data processing tasks daily and improving overall operational efficiency by 50%
• Developed a user-centric FastAPI interface, seamlessly integrating it with an intuitive API Server to facilitate efficient data ingestion and retrieval for cross-functional users accessing the Datalake.
• Technologies and Tools: Trino, Python, Apache Airflow, MongoDB, Apache Kafka, Apache Nifi, FastAPI, Trino, MinIO, Docker

Lead Management System:

• Designed and implemented a FastAPI-based REST endpoint to accept lead conversations and generate AI-powered summaries.
• Integrated MongoDB (Replica Set with TLS authentication) for secure storage of conversation data.
• Built a single unified API to store raw conversation data and return LLM-generated summaries.
• Integrated Ollama with the qwen 2.5 model for local large language model (LLM) summarization.
• Developed structured prompt engineering logic to extract customer intent, key discussion points, and next steps.

Payment orchestration platform:

• Built a real-time payment gateway selection system that routes transactions based on cost, latency, and success rate, reducing failures by 20%.
• Developed REST APIs using FastAPI to handle live transaction requests smoothly and ensure reliable processing.
• Integrated multiple payment gateways and implemented routing logic to automatically choose the best gateway for each transaction.
• Created synthetic datasets to simulate real payment scenarios, helping validate business logic and improve testing accuracy.
• Deployed the application on a Kubernetes cluster to ensure high availability, scalability, and minimal downtime.