Software / AI Engineer
Hi, I'm Krisha.
I build intelligent systems end to end — LLM workflows, RAG pipelines, and the backend infrastructure that keeps them running in production.
About me
I’m a Software Engineer at A.P. Moller Maersk, where I build things end-to-end — from database schema and backend API to frontend and CI/CD pipeline. My work runs in production with real users, SSO authentication, and real stakes.
I specialize in GenAI and backend systems — LLM workflows, RAG pipelines, async APIs, and data engineering with Python and SQL. I also have research published in Scopus-indexed journals on multimodal RAG and ML-based malware detection.
Work
EPICLabs EM Copilot
AI / AgenticLLM-powered analytics copilot for Maersk's Empty Model ops team. Answers natural-language questions about empty container costs by querying a Databricks-backed data warehouse and synthesizing answers via GPT-4o — reducing analysis time from hours to seconds.
Built: Async job engine with polling and cancellation, 4-workflow LangGraph orchestration (gradual change, missing cost, abrupt change, RAG docs), intent router with entity extraction, structured logging, and GitHub Actions CI/CD to AKS.
- FastAPI
- LangGraph
- GPT-4o
- React
- PostgreSQL
- Docker
- AKS
FapiaoFlow
Full-Stack / Internal ToolException review tool for Maersk's China Cash Application team. Surfaces ~750 Fapiao invoice failures daily from a 5,000+ req/day Databricks pipeline across 10 China offices — letting ops teams review, action, and sync decisions back to Databricks automatically.
Built: Two-table ownership database schema, async FastAPI backend with SQLAlchemy, Azure AD JWT auth, Databricks push/pull sync with audit logging, and a Vue 3 + Pinia frontend.
- FastAPI
- Vue 3
- PostgreSQL
- Databricks
- Azure AD
- Docker
My projects
PromptEase
Multimodal RAG chatbot for multilingual product support using Gemini and ColPali-based retrieval across 170 technical documents. Published in IAJIT (Scopus-indexed).
- FastAPI
- LangChain
- Gemini
- ColPali
- Vector Search
CropPal
Crop identification and growth-stage detection system using CNNs and transfer learning. Achieved 92%+ classification accuracy across 14,000+ labeled images.
- Python
- CNN
- EfficientNet
- Transfer Learning
- Flask
My skills
- Python
- SQL
- LangChain
- LangGraph
- RAG
- LLM Application Development
- FastAPI
- React
- REST APIs
- PostgreSQL
- PySpark
- Apache Kafka
- Delta Lake
- Databricks
- Azure
- Docker
- Git
- Power BI
- Pandas
- Machine Learning
Research Publications
Optimising Multimodal RAG Systems for Multilingual Product Support
IAJIT (Scopus-indexed)
Evaluated 5 RAG architectures across 17 configurations for multilingual product support. A ColPali-inspired multimodal approach achieved 94% Top-1 retrieval accuracy.
Malware Detection and Classification Using Machine Learning Techniques: An Implementation Study
IJCDS (CiteScore 2.3, Q3)
Developed ML-based detection models for polymorphic URLs and PE malware, achieving 97.85% and 99.4% accuracy using TF-IDF/Logistic Regression and Random Forest with oversampling techniques.
Contact me
Please contact me directly at krishajoshi.work@gmail.com or through this form.