BBAbout Bhuwan
I help businesses and teams turn complex AI challenges into scalable, production-ready solutions. With 5+ years of experience in ML/AI, LLMs, and MLOps, I specialize in optimizing workflows, cutting operational costs, and boosting model performance. My edge lies in combining deep technical expertise with hands-on cloud experience, particularly on AWS, to design, deploy, and maintain high-impact AI systems. I’ve delivered projects ranging from LLM-powered medical transcription and recommendation engines to document ranking systems and predictive analytics. Whether it’s building end-to-end ML pipelines, fine-tuning large language models, or architecting cost-efficient cloud infrastructures, I bring a track record of delivering solutions that are not only accurate but also fast, reliable, and easy to scale.
English
Native or bilingual
Can work on-site
London (up to 50km)
Experience
- Tech42ML EngineerMay 2025 - May 2026 (1 year)
- Delivered AWS Bedrock-powered AI solutions for multiple enterprise clients across healthcare, logistics, travel, and media industries
- Architected end-to-end LLM pipelines, agentic systems, and production ML infrastructure
- Built a multimodal semantic search system (Facer) using Claude and Nova embeddings, cutting embedding costs 50% via batch inference on 500K–600K items
- Designed a serverless procurement extraction pipeline (American AI Logistics), cutting inference cost 90% through model benchmarking and prompt engineering
- Shipped a production real-time speech service (Opencall) with sub-1s latency and cut autoscaling detection time 6x
- Deployed a multi-tenant LLM-powered asset health scoring pipeline (Service Insight) with full MLOps infrastructure and zero-secrets-in-code architecture
- Architected a production LangGraph agentic system (COG AI) for healthcare vitals monitoring, including PHI encryption and parallelized data pipelines
- Built and deployed a LangChain-based travel booking agent (bTd) handling real flight transactions end to end, with content safety guardrails
- Productionized an AI recipe management agent (Bacon) with automated PDF-to-recipe digitization and LLM spend controls
- Triumph Technology SolutionsML EngineerSeptember 2022 - August 2023 (11 months)Kathmandu 44600, Nepal• Automated transcription of doctor-patient conversations using Amazon Transcribe Medical, improv ing documentation efficiency.• Extracted key medical entities (medications, diagnoses, symptoms) using Amazon Comprehend Medical, enhancing structured data retrieval.• Generated structured SOAP notes using LLMs (e.g., ChatGPT, Stable Vicuna), reducing man ual effort by 50%.7Cups• Translated client requirements into actionable ML solutions through discovery sessions.• Optimized the platform's recommendation system using AWS Personalize, enhancing content relevance and user engagement.• Analyzed recommendation engine performance through data insights, identifying improvements in post relevancy.• Developed an interactive Amazon QuickSight dashboard to visualize key engagement metrics, improving decision-making.Others: Traive, Zetane, LucidAct, Cellereye• Architected scalable, fault-tolerant AWS-based ML infrastructures, optimizing performance and reliability.• Conducted in-depth latency analysis to estimate request delays during cloud migration, ensuring seamless transition.• Developed comprehensive AWS cost estimations using the AWS Pricing Calculator, enabling clients to budget efficiently.• Automated nursing summary generation at LucidAct using LLMs (ChatGPT, Stable Vicuna), reducing manual workload.• Implemented LLM-powered explainability models at Traive, improving financial ML decision transparency.Tools: Lambda, S3, EC2, Amazon Personalize, Amazon ECR, Sagemaker, QuickSight, Large Language Models, ChatGPT, Python, Stable Vicuna, Falcon 40B, HuggingFace, Pytorch, Pandas, Data Analysis, Cloudwatch, Glue, ETL, VPC, Linux, Unix, Exploratory Data Analysis
- UpworkML EngineerSeptember 2017 - December 2023 (6 years and 3 months)• Developed ML solutions for diverse clients in finance, healthcare, and manufacturing, delivering scalable and production-ready models.• Designed and optimized stock prediction models for the US market, integrating sentiment clas sification to reduce false positives.• Built OCR-based text extraction pipelines and performed information retrieval from unstruc tured image-based documents.• Engineered an audio-based equipment defect detection system using signal processing and deep learning, improving predictive maintenance.• Employed LIME and SHAP to enhance explainability of ML model predictions, increasing trans parency and trust in AI-driven decisions.• Mentored a high school student in Python and machine learning, fostering foundational knowledge in AI.
Recommendations
Be the first to recommend Bhuwan
Help this freelancer shine by sharing your experience working together.
These freelancer profiles also match your criteria
Agatha Frydrych
Backend Java Software Engineer
4.7
(3)
2
Baptiste Duhen
Fullstack developer
4.6
(4)
5
Amed Hamou
Senior Lead Developer
4
(2)
7
Audrey Champion
Web developer
4.3
(3)
4
Education
- Masters of Advanced Computer ScienceUniversity of Hertfordshire2024Masters of Advanced Computer Science
- Bachelor of Computer EngineeringInstitute of Engineering, Tribhuvan University2018Bachelor of Computer Engineering