Anjana Ramachandran

Multimodal Integration: Built comprehensive ML-based breast cancer subtype classifier integrating 4 distinct modalities: genomic sequence reads, gene expression profiles, radiomic features, and histopathologic images from 112 patient samples from HCG Cancer Hospital, Bangalore.

Modality-Specific Pipelines: Engineered individual ML pipelines - variant calling using GATK for genomics (87% accuracy), feature extraction with Pyradiomics for radiomics (92% accuracy), deep CNN with MONAI for histopathology (89% accuracy), and expression profiling for transcriptomics (91% accuracy).

Ensemble Innovation: Implemented sophisticated weighted-voting ensemble algorithm that improved subtype classification accuracy to 93.5%, achieving 14% performance gain over single-modality approaches, specifically optimized for Indian patient cohort characteristics.

4-Phase Workflow: Designed and deployed comprehensive agentic AI workflow with MCP (Multi-Channel Protocol) server integration encompassing lead prioritization, qualification, automated proposal generation, and intelligent follow-up automation across the entire sales cycle.

Multi-MCP Integration: Integrated multiple MCP servers including CRM integration, document processing, communication management, and data enrichment services, enabling seamless context-passing between agents, real-time notifications, and personalized proposal generation.

Conversion Optimization: Improved lead conversion efficiency by 35% through streamlined BANT (Budget, Authority, Need, Timeline) scoring algorithms, automated proposal delivery systems, and sophisticated drip campaign automation across the complete sales funnel.

Network Construction: Engineered sophisticated dual-target pathway model achieving 95% validation accuracy through novel edge-weight formula combining gene overlap coefficients, differential expression significance, and pathway importance metrics across 40 biological pathways with 589 interaction edges.

Omics Integration: Executed comprehensive differential gene expression analysis using DESeq2 on multi-omics datasets, successfully identifying 89 common genes with significant therapeutic roles in both T2DM and PDAC, validated through multiple independent datasets.

Algorithm Development: Applied influence propagation algorithms and Linear Threshold models to pinpoint critical pathways for dual-drug targeting, enhancing predictive cancer treatment models through Boolean logic implementation and network topology analysis.

Advanced NLP Pipeline: Developed comprehensive question-answer generation system using spaCy for binary classification to identify potential answer keywords, enhanced with Part-of-Speech tagging and Named Entity Recognition for improved accuracy.

MCQ Generation: Implemented cloze-style methodology facilitating automated Multiple Choice Question generation directly from text segments, utilizing word embeddings and cosine similarity algorithms to create semantically plausible distractors.

Classification Excellence: Applied Gaussian Naive Bayes for robust word classification achieving 99.2% accuracy, demonstrating successful integration of AI with software engineering methodologies for educational technology applications.

Multi-Algorithm Implementation: Developed and implemented comprehensive suite of optimization algorithms including Particle Swarm Optimization (PSO), Ant Colony Optimization (ACO), Genetic Algorithm (GA), and Cultural Algorithm for complex problem-solving and optimization tasks.

Sequence Alignment Innovation: Engineered C++ Affine Gap Penalty Model for highly efficient biological sequence alignment using dynamic programming principles, incorporating adaptive parameter tuning and dynamic velocity adjustments.

Hybrid Optimization: Applied hybrid cultural evolution techniques combining multiple optimization strategies for enhanced convergence rates and solution quality in computational biology applications.