Aditi Jaiswal

Developing product features using LLMs such as OpenAI, LlamaIndex, Llama2, Claude3, enhancing their capabilities for user specific tasks. Designed and implemented a next question prediction feature, enabling the model to anticipate and suggest follow-up questions. Created summarization prompts to enable the model to generate concise and informative summaries of lengthy text. Engineered support for various document types and developed question-answering systems. Enhanced document reading capability using table-OCR and GPT4 Vision API. Enhancing RAG specific tasks using AWS-Bedrock.

Leveraging advanced NLP algorithms to build custom models for entity extraction, concept recognition, relation extraction, summarization, textual classification and clustering. Implemented Parts of Speech Tagging (POS) using Spacy for identification of search terms. Worked on Language Detection and Multiclass vertical prediction using CNN, LSTM and State of Art algorithms. Developing Spelling correction module using Elasticsearch query, Character embeddings, BERT etc. Fetching and updating data using Elasticsearch and Kibana. Studying user behavior for spellchecks and next sentence prediction.

Research and implementation in Dialogue Systems, Speech Recognition, Dynamic Knowledge graph construction and other NLP and Computer Vision models. Defining data augmentation pipelines. Training models and tuning their hyperparameters. Analyzing the errors of the model and designing strategies to overcome them. Deploying the models to production.

Responsible for the application of the State of Art Conversation Driven Development techniques to design conversations. Built and evaluated NLP pipeline. Integrated Chatbots with Management Dashboard and CRMs. Lead the Test-Driven Development & Agile Methodologies.

Developed and maintained robust data processing pipelines and reproducible modelling pipelines. Extracted real-time data from Quandl API and applied Forecasting algorithms like ARIMA, SARIMA, ETS etc. Designed pipelines in Azure for workflows. Handled Time Series Forecasting using standard Regression algorithms like Linear Regression, Gradient Boosted Decision Trees, LSTM, GRU, and Transfer Learning. Used Naive Bayes, KNN, Linear SVM and LSTM and compared their performances.

Extracted Raw Data using complex SQL queries from GeoEnhance API. Cleaned the Data using Excel and provided analysis based on their locations. Helped the company in generating the Point of Interests for Consumer Analytics.