Mohan Vamsi Adluru

M.Sc. Applied Computational Life Sciences — GPA: 4.6 / 6.0

• Clinical NLP & LLM Fine-tuning: Replicated a hospital course summarization pipeline benchmarking Clinical-T5, LLaMA2-13B, and GPT-4 on the MIMIC-IV-BHC dataset; QLoRA fine-tuning improved BERT F1 by +6.3% on Clinical-T5 (0.584 → 0.647) and +7.1% on LLaMA2-13B (0.612 → 0.683), with LLaMA2 outperforming the specialised clinical model post-tuning.
• Knowledge Graph Extraction: Fine-tuned Mistral-7B on the NewsKG21 dataset for subject–predicate–object triple extraction; evaluated three strategies (gold standard, spaCy-filtered, combined), achieving best F1 of 0.211 with the combined approach using Alpaca-style prompting and LoRA on HPC infrastructure.
• Biomedical Pipelines & Computational Optimization: Built an NLP pipeline over Medline XML and ChEBI ontology for chemical NER and entity resolution; developed a genetic algorithm optimising insulin dosage, meal intake, and exercise timing for Type 2 diabetes patients targeting a 100 mg/dL glucose level within a safe clinical range (80–130 mg/dL).

B.E. Computer Science and Engineering — GPA: 7.96 / 10

• Built a real-time Remote E-Proctoring System using NLP for behavioural anomaly detection and OpenCV for computer-vision-based malpractice identification during online exams, integrated with Amazon S3 for scalable evidence storage.

MSc Thesis Researcher

• Designing an evidence-based decision support system that helps clinicians and nutritionists interpret gut microbiome profiles from stool sequencing in the context of inflammatory bowel disease, by grounding treatment decisions in structured knowledge extracted from biomedical literature (NDA-covered project).
• Building a modular, reproducible pipeline that links sequencing-derived microbiome readouts with curated evidence on diet, interventions, and disease mechanisms, so that AI systems can provide traceable, literature-backed recommendations rather than opaque model outputs.

Mobile Application Developer (Startup Collaboration)

• Built the full-stack mobile application in React Native to support a personalised dietary recommendation engine for a food-tech startup, enabling users with allergens and dietary restrictions to discover safe, curated restaurant options and menus.
• Designed the data schema in Supabase/PostgreSQL linking customer dietary profiles (allergens, intolerances, preferences) with restaurant and menu data, laying the structured foundation for downstream personalised recommendations; implemented secure multi-provider authentication and Row Level Security policies throughout.

Associate Product Developer

• Took end-to-end ownership of 9 of 16 Power BI visuals shipped to production, driving each from initial concept through design, implementation, and release, directly contributing to Microsoft AppSource Certification and 92% enterprise client adoption across the product suite.
• Reduced front-end rendering latency by 35% through algorithmic refactoring and optimised data-binding strategies using D3.js, Highcharts, and AG Grid; resolved 300+ production issues via systematic root-cause analysis across large enterprise datasets and cross-browser environments.
• Contributed to agile sprint planning, peer code reviews, and technical documentation within a cross-functional product team, maintaining a consistent quarterly release cadence.

Undergraduate Research Assistant

• Developed a critical-state-detection mechanism for deep RL agents under adversarial attack; attacking <1% of states reduced agent performance by 40%+, while the proposed long-term impact classifier reduced compute time by 80.3% over prior methods, published at IEEE ICMLA 2021.
• Built a two-stage flight delay prediction pipeline using SMOTE, Random Forest, and XGBoost for delay classification and duration estimation (R²=0.95, RMSE=16.42, MAE=11.45).

ZHAW × Roche Hackathon 2025 | 2nd Place Winner — Sep 2025

• Prototyped an AI-driven remote patient monitoring solution for chronic disease management, integrating wearable data analytics and predictive modelling aligned with Roche's Hospital at Home initiative, securing 2nd place across all competing teams.

Sustainability RAG System | SDSC Workshop - Canton of Zurich — Feb 2026

• Participated in an SDSC-facilitated industry workshop building a RAG system for EU CSRD compliance querying; hands-on experimentation with hybrid retrieval strategies (BM25, dense vector, HyDE, LLM reranking), multi-agent architectures (RAG Agent, ReAct ToolAgent, LLM Router), and evaluation frameworks (RAGAS, MRR, NDCG, Hit Rate).

HealthTech Summer School | Biodesign & Digital Health Innovation — Jul 2025

• Acted as CTO in a multi-disciplinary team; led SOTA analysis of AI diagnostic technologies, conducted patent landscape and gap analysis, and assessed regulatory pathways, from clinical immersion to final industry pitch.

Innosuisse Business Concept Programme | Startup Campus Switzerland — Feb-May 2025

• Completed a 12-week Innosuisse-recognised business concept acceleration programme as part of the founding team of KIM; led technical feasibility scoping, evaluated sensor-based and computer vision approaches for species identification, and pitched the solution architecture before a jury.

SIB Swiss Institute of Bioinformatics | LLMs for Biodata Exploration — Sep 2025

• Built a SPARQL query generation assistant over federated biomedical endpoints (UniProt, Bgee, OMA) using LangChain and Qdrant.