Precious

Precious Studio is a global UX agency headquartered in Austin, TX. We specialize in creating bold brands, designing catchy websites, and developing digital products that are trendsetting. With a clientele of over 150 development and marketing teams, Precious is recognized as a Top UX Company in the US by Clutch in 2023.

Check out verified client review
https://clutch.co/profile/precious

Principal UX Consultant

Pirate Squad

PirateSquad is a Design-as-a-subscription service. It is the one subscription that you need for all your product and marketing needs.

Co-Founder

Various Startups

Mentor/Advisor

Thiken

Co-founder/Head of Design

OATI

Software Developer

Sundog

User Experience Architect Intern

North Dakota State University

UI/UX Designer (IT Division)

Principal UX Consultant @ Precious Studio

👋 Hey there! I'm Pavi. 

We're all about putting our heads together to create unique designs, developing forward-thinking UX strategies, and diving deep into User Research. 

Creating a distinctive brand and engaging user experiences can differentiate success and failure. As your partner, I aim to provide solutions that enhance your digital presence and contribute to your business growth.

✔️ Whom I Assist
I work with startups and agencies aiming to create a unique digital identity. If you're keen on harnessing the power of user-centric design to make an impact, I'm here to help.

✔️ Problems I Solve
Standing out and offering a seamless user experience can be challenging in a cluttered digital world. My solutions are designed to navigate these challenges, leveraging innovative, UX-driven design tailored to your unique needs.

✔️ How I Add Value
With my consultancy, recognized as a top-rated UX agency in the US by Clutch in 2023, I help businesses transform their digital narratives through:

 - Bold Branding: Crafting unique identities that resonate with your audience.
 - Award-winning Apps: Creating mobile and enterprise apps that have earned recognition for design excellence.
 - High-converting Websites: Building user-friendly websites that boost conversions.
 - Advanced Prototyping: Exploring advanced prototyping and motion design to bring your brand to life.
 - Scalable Systems: Developing design systems that ensure consistency and streamline processes.
- Engaging Marketing: Boosting your brand visibility with strategic marketing efforts.

🏆 Proven Track Record
My work has earned significant recognition, such as the Best App Design Award by DesignRush in 2023 and a top trending app on the Google Play Store. I've had the privilege of collaborating with big brands like Google and Mitsubishi and over 100+ innovative startups.

✔️ The Distinct Difference
What sets me apart? It's simple. I listen, understand, and then deliver. I focus on understanding your brand and your audience deeply. From there, I build user-centric, tailor-made solutions that scale with your business and resonate with your audience. That's my promise to you.

📩 Let's Collaborate
Would you be ready to amplify your brand and digital offerings? Reach out directly at pavithra@precious.team or send me a message here on LinkedIn. Together, we can redefine your digital presence.

Pavithra Lamahewa

Wipro

Director, Data Analytics & AI

Cognizant

Technology strategy consulting  and delivery of Data and AI/ML projects for Life Sciences and Financial Services firms in the DACH region.

Principal AI/Analytics Architect

Darwin Digital

Darwin Edge (https://darwinedge.com/) supports companies to exploit the true potential of Edge AI computing. 

Edge AI enables intelligent consumer focused solutions to be deployed on edge devices, reducing latency, improving energy efficiency and providing strong privacy guarantees.

We have proven experience in delivering Visual, Location and Analytical solutions at the edge for diverse industries, such as Healthcare, Automotive, Manufacturing, Retail and Energy. 

You will be able to leverage our team's expertise in AI, Data and Hardware Engineering to move your AI prototypes to products, ensuring that your applications are open, scalable, cost-effective, and deployed in a responsible fashion. 

Chief Technology Officer

Philip Morris International

Led Artificial Intelligence (AI) strategy execution at PMI: 
•	Formation of the AI Center of Excellence
•	Delivered AI solutions to different PMI functional units in the fields of Predictive Analytics, Forecasting, NLP and Image Recognition. 
•	Setup an innovation pipeline (process and governance) for AI related use-cases, starting from demand gathering to PoCs to productization.
•	Govern and contribute towards Analytics Strategy, Technology Roadmap and Architecture of PMI’s (AWS based) Big Data platform.  
•	Establish a partner ecosystem consisting of cloud and service providers, universities and startups.
•	IPR generation and protection, including advancing the state-of-the-art with scientific publications and Open Source contributions to AI research frameworks. 

Established PMI’s Open Source Governance Office. Activities include Open Source license management, scanning tool roll-out and managing Open Source contributions.

Principal AI/Analytics Enterprise Architect

Swisscom

Design and develop the architecture of Swisscom’s Analytics platforms.

Big Data stack:
- Architect Data Science delivery pipeline with technology and tool selection.
- Develop privacy guidelines for the Big Data platform.

Data Warehouse (DWH): 
- Provide data modeling guidelines for the DWH platform. 
- Co-ordinate and manage the activities of diverse analytics teams involving marketers to solution developers, for new marketing service requests.

BI Enterprise Architect

Iprova Sàrl

Iprova is transforming the way new products and services are invented by using data and AI/ML technologies to enable customers to have a sustainable competitive advantage.

- Led the design and development of Iprova's software platform. 
- Filed 25+ patents for a number of Fortune 500 companies.

Technology Leader

Nokia Research

Research on Geo-location based Services, Security & Privacy

20+ Patents and scientific publications in related fields.

Senior Researcher

Research on Distributed secure protocols for SAP Supply Chain Management (SCM) systems - as part of the EU project SecureSCM.

Researcher

Oracle Corp.

Prototype development and consulting focusing on Oracle 9iAS (J2EE
Application Server).

Technical Consultant

Infosys Technologies Limited

Software development and design on J2EE/C++/Web technologies.

Software Engineer

Debmalya Biswas has over 20 years of experience working on AI/ML, Big Data Analytics, Privacy & Security related topics; most recently as Director, Data Analytics & AI at Wipro, Switzerland. 

He has previously worked for leading technology companies, e.g., Nokia Research, SAP Research, Oracle, Swisscom, and Swiss startups Iprova and Darwin Digital. 

He has authored 50+ patents, 30+ scientific articles, is a regular speaker at leading AI/ML conferences and is a member of the AI4EU Evaluation Committee. He holds a PhD in Computer Science from INRIA, France.

Debmalya Biswas

#Human-in-the-Loop (HITL) strategy for #AgenticAI
Given the unreliability of current #LLMs / reasoning models, studies have shown that a collaborative human-assisted model works much better in terms of task fulfillment today.

For #autonomous agents, HITL is also the promise that we make to our #Legal & #Compliance teams to get their go-ahead :) But do we really have them effectively embedded in your agentic #workflows?

Given this, the proposal is to integrate humans as a first-class citizen in the agentic #lifecycle, and not only as a supervisor / reviewer. This also implies that appropriate UI / #UX needs to be designed to allow human intervention for the right task at the right time. (Similar to agentic state #checkpointing, the frequency and format used to solicit human feedback matters.)

Below is a list of the key human #intervention points to plan for:
- Co-plan: Validate & #plan ensuring that the generated plan corresponds to the user intent.
- Co-execute: Users can intermittently pause the #execution and give feedback, if the agent’s / tool's response does not comply with the assigned task. 
- Co-comply: Users can mark the critical and irreversible tasks (e.g., payments), and ensure that right #guardrails have been applied compliant with enterprise policies - before approving the task.
- Co-memorize: Refine #memory, reviewing key long-term memory concepts, optimizing storage, ensuring reusability.

This is complemented by a continuous improvement module that learns from historical #interactions to optimize future human interventions.

(views are my own)

AI @ UBS | x- Nokia, SAP, Oracle | 2x Startups | 50+ patents | PhD - INRIA

1. Introduction to Agentic AI The discussion around ChatGPT (in general, generative AI), has now evolved into agentic AI. While ChatGPT is primarily a

Open article: Human-in-the-Loop Strategy for Agentic AI by Debmalya Biswas • 9 min read

Human-in-the-Loop Strategy for Agentic AI

It was an enriching and humbling experience to present our #AgenticAI work at the J.P. Morgan Global #MachineLearning Conference in Paris yesterday.

The grand ambience and the collective intelligence of the attendees  in the room set the stage for many memorable conversations. As we move towards maturing our Agentic AI platform, I had a chance to talk about some of our upcoming work around observability, long-term #memory, evaluation, and #guardrails.

Some takeways from the event:
- People are still skeptical about the value of neural networks for time-series #forecasting.
- Transformers + Classical Automated Planning techniques seem to be the way to go towards overcoming the #reasoning limitations of #LLMs. Knowledge graphs, #CausalAI, Memory are key here.

- Evaluating (non-deterministic) #agents remains notoriously difficult. People are using a mix of manual and LLM-as-a-#Judge approaches. However, they are far from comprehensive and 'complete'. 
- Generating test data synthetically can be an alternative. However, generating #synthetic test data remains challenging due to the need to balance closeness to existing data, and differentiation to cover edge cases - the same challenge in choosing between exploitation and exploration in #ReinforcementLearning.

- Most #GenAI use-cases today are targeted towards processing (large) documents, basically #unstructured data, or generating insights / reports by collating information from multiple data sources / document repos.

Overall, there was (hopeful) optimism in the room that we are in the early stages of this journey, so build things in a way that can #adapt to change - rather than trying to build a future proof platform!

(views are my own)

taplio

The Battle

Pavithra Lamahewa

Debmalya Biswas

Followers

Posts

Engagements

Likes

Performance ScoreDebmalya Biswas's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Pavithra Lamahewa

Debmalya Biswas

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Debmalya Biswas's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreDebmalya Biswas's LinkedIn Performance