Pranab Das

Case Studies

• Revenue Increase for Airlines
• Deep Learning for Pharma
• Payment Engine for Major Bank
• Call Center Call Moderation

Problem: Drop In Air Seat Miles Revenue for Major US Airline

In 1999, a major US airline faced a concerning drop in their Air Seat Miles revenue due to operational inefficiencies in their communication systems. Their new Automated Call Distributors (ACDs) were failing to route emergency calls from different flight departments (e.g., cockpit, cabin) to the nearest or appropriate Aircraft Maintenance Call Centers. Each department had its own dedicated 1-800 emergency number, enabling urgent communication; but incorrect routing caused widespread delays and cancellations---incurring a million-dollar daily loss.They sought my expertise after the provider of the communication switches was clueless. Within hours, I implemented a band-aid solution to stabilize the calls temporarily. A long-term architecture was designed and implemented over two years, still functional during 9/11.Technical Approach & Solution:The system leveraged the Public Switched Telephone Network (PSTN) and routed calls via virtual numbers managed by Intelligent Network (IN) logic, which was the immediate recourse for all virtual numbers, to ensure efficient destination routing. Alongside voice, the system transmitted essential data (flight details, department ID, etc.), directing calls to the most qualified personnel.One key challenge was equipment incompatibility between the newly installed switches and the agent systems. To address this, I created a low-level protocol translator to bridge the technical gap.Additionally, the incredibly high availability requirements (approaching six sigma) led me to select a Publish-Subscribe model using Message-Oriented Middleware (MOM), ensuring seamless communication between all stakeholders.

Problem: Drive Personalized Medical Suggestions for Pharma

A prominent pharmaceutical firm required a recommendation engine to drive personalized medication suggestions on their website. The core goal was to recommend their OTC allergy medication based on user profiles, time of year, and local pollen levels.I architected a solution using IBM WebSphere integrated with Blue Martini, one of the few recommendation engines available at the time.Solution & Technical Stack:- Recommendation Engine: Blue Martini
- Web Platform: IBM WebSphere
- Database Integration: Real-time pollen data.The project marked an innovative step toward personalized health recommendations – a rarity at that time.

Problem: Secure Payments Engine for Major Banking Client

In response to the growing need for electronic payment systems, a US retail bank entrusted me with designing a system that facilitated ACH (Automated Clearing House) transactions alongside printed checks. The complexity lay in addressing the necessary legal compliance while keeping security airtight.Innovation & Solution:This project pioneered the use of Link Level Encryption (LLE), which allowed even insider threat mitigation, using 128-bit encryption.

Problem: Cuss Word Detection for Call Centers

A productivity tool for call center training required maintenance and refinement. The tool flagged inappropriate language or sentiments in employee speech. Upon conducting a technical audit, I found that the existing implementation was inefficient and underperforming, mainly due to improper use of Java programming constructs.Solution & Innovation- Optimized the performance by restructuring how variables were scoped (switching from class-level to instance-level where needed).Ensured the product met the necessary performance benchmarks to compete with equivalent call training products.