Parin Acharya@parin_acharya89

Trained LLMs to automate financial analysis using parallel processing to gainin a competitive insight edge on 10k NYSE stocks Upscaled knowledge base by extracting sentiment from market firehose ingesting 100 articles/sec from real-time news API

Real-time image feed from HuggingFace stable diffusion model on serverless a10g GPU, reducing cold-start to <1sec Built recommendation system for personalized feeds and enhanced user content management with prompt histories

Used Next.js and Cheerio to build a web URL crawler for citation-based responses, inspired by perplexity.ai’s $9b valuation Handled open-source LLM rate limits using caching on Redis, reducing latency by 5x and improving scalability

Why WorkWiseAI? Large modern businesses and fast-growing startups often struggle with operational inefficiencies hidden within their workflows. Traditional process audits can be time-consuming, subjective, and costly. WorkwiseAI automates this process, delivering real-time, AI-driven workflow analysis that enhances decision-making. Use cases of IBM Services have seen success, including Vodafone, which reduced testing timelines to 1 minute and Artefact, which uses a portfolio of personas represented by AI identities, allowing professionals to reveal crucial insights from customer behavior. Built on IBM’s open source granite-3.1-8b-instruct powered by IBM AgentLab, WorkwiseAI provides businesses with deep workflow insights, data-driven efficiency scoring, and AI-generated optimization strategies. At the core of WorkwiseAI is a custom 'S3' Agent Pipeline that orchestrates three AI-powered agents and IBM Vector Index document grounding to streamline workflow analysis: Summarizer Agent: Breaks down complex business processes into distinct, numbered steps, identifying inefficiencies based on structured business context data. Scorer Agent: Critically evaluates each workflow step, assigning efficiency scores using industry-relevant metrics and a custom scoring model. Suggester Agent: Leverages tools to access external knowledge sources such as Wikipedia, DuckDuckGo, and retrieval-augmented generation (RAG) queries to provide actionable improvements for each workflow step. The prototype also enables interactive Workflow Visualization: clean structured insights from large data through intuitive Streamlit dashboard widgets. Multi-Language Chatbot Interface: Allows for seamless interaction with WorkwiseAI across different business environments. By harnessing IBM’s advanced AI ecosystem, WorkwiseAI empowers organizations to transform workflow management, ensuring agility, competitiveness, and sustainable growth.