Inpharmativ Consulting
Ancaster, ON
Healthcare data and life sciences consulting
• Amazon S3 • AWS Lambda • Amazon API Gateway • Amazon Bedrock • Amazon RDS for PostgreSQL • Amazon OpenSearch Service • AWS Identity and Access Management • Amazon CloudWatch
Inpharmativ Consulting builds lists of relevant physicians using public research sources. The company wanted to automate and scale this work with generative AI, while keeping costs and time to value tight. Avahi designed and implemented a production ready data and retrieval system on AWS that automates entity extraction, powers a RAG based search assistant, and exposes results through secure APIs. The engagement was structured to fit a six week engineering window, funded through AWS GenAI Prod credits.
Inpharmativ Consulting operates in healthcare data and life sciences, helping clients identify physicians and researchers based on publications, conference activity, and related public signals across Canada.
Manual physician discovery did not scale with growing data sources or query complexity. Analysts had to sift through research papers, abstracts, and author profiles, which slowed time to answer and increased the risk of inconsistent results.
Inpharmativ needed automated extraction of people, institutions, and medical topics, plus a way to connect physicians to conditions with semantic relevance, then return results programmatically. Without a solution, the team would continue to spend large amounts of analyst time on repetitive search and curation work, and turnaround times would lengthen as data volume grew.
Any automation also had to respect a controlled list of public sources, including PubMed, Google Scholar, conference websites, and journal databases.
AWS provided managed building blocks for generative AI and data pipelines, including Amazon Bedrock for model access, serverless services for ingestion and APIs, and managed storage for structured and vector data. Using these services helped the team deliver quickly without heavy platform operations.
The project also aligned with AWS GenAI Prod funding, which contained cost and supported the six week delivery window.
Avahi is an AWS Premier Tier Services Partner with repeatable patterns for GenAI search, entity extraction, and production deployment. The team committed to rapid iteration, frequent demos, and customer validation in three day cycles, and to delivering within the client timeline.
Avahi deploys into the customer AWS account, the customer retains ownership of code, models, APIs, and outputs, which matched Inpharmativ governance requirements.
Avahi implemented an end-to-end pipeline that ingests public research data, enriches it with NLP, and serves ranked physician results through an API.
Data ingestion and preprocessing. Sources included PubMed, Google Scholar, conference websites, and journal databases. Enhanced web scraping extracted structured and unstructured text from papers, abstracts, and author profiles. The pipeline normalized text, removed duplicates, and prepared multilingual content for downstream NLP.
Entity extraction and enrichment. Using Amazon Bedrock hosted models, the system applied named entity recognition to capture physician names, affiliations, research topics, and medical conditions. Embeddings were generated to compute semantic similarity between a physician body of work and user specified conditions. This produced machine readable relationships that map physicians to medical focus areas.
Storage and retrieval. Physician relationships are stored in Amazon RDS for PostgreSQL. Embeddings are indexed in a vector store, Amazon OpenSearch Service or ChromaDB. Retrieval supports semantic search and a RAG workflow.
Smart assistant and APIs. A retrieval optimized knowledge base feeds an LLM through Amazon Bedrock that assembles the final answer. Results are served through Amazon API Gateway and AWS Lambda, which lets Inpharmativ integrate directly with internal applications.
Operations and security. Data staging uses Amazon S3. Logging and metrics use Amazon CloudWatch. Access is governed with AWS Identity and Access Management. Avahi delivered documentation, deployment scripts, and runbooks limited to the features created in this project.
The engagement met Inpharmativ needs for a production ready pipeline and API, while keeping scope and cost inside a tightly managed window. The result is an AI assisted discovery experience that surfaces relevant physicians by condition and research focus, backed by documented runbooks and clear ownership in the customer AWS account.
Documented metrics:
Founder, Bravo Foxtrot
