You are on page 1of 19

Think Data, Think Sigmoid

Jan 20th, 2022


A Leader In Data Engineering And AI Solutions

Backed by
Work with Customer
500+
25+ Data Pipelines
Satisfaction Score
Fortune 500 firms Built 97.6%

100+ 50+ PB 100% 400+


Analytics models Data handled Employees from Data
operationalized daily Tier 1 Scientists &
Engineers

Offices
New York Dallas San Francisco Lima Bangalore Amsterdam
2
Recognized For Technology Capability And Innovation

2021 Open-Source Data


Solution Provider of
2021 the Year

2020 2021

Recognized in
FORRESTER
Now Tech: AI Consultancies, Q1, 2021 report

2020 2017 2014

Innovation in Digital
Transformation

3
World’s Largest Data Producers Trust Us To Deliver Business Impact

We build, process and manage the world’s largest data platforms


Top-3 retailer | Top-3 global investment bank | Leading AdTech provider | Top-3 CPG company in LATAM

and many more F1000 companies across retail, CPG, BFSI, pharma, advertising, QSR and high-tech industries

4
Integrated D&A offerings yield successful business outcomes

Building end-to-end engineered AI/ML solutions

CONSULTING DATA SCIENCE DATA ENGINEERING MANAGED SERVICES

• Data Strategy Development • Digital Transformation • Building ELT/ETL Pipelines • Data Lifecycle Management

• Business Problem Exploration • Data Modelling • Implementation of EDW • 24/7 Monitoring and Support

• Consultation Workshop • Machine Learning • MLOps • Infrastructure Automation

• Technology Comparison and • Artificial Intelligence • Data Quality & Governance • DevOps Adoption
Recommendations
• Deep Learning • Cloud Data Warehouse • Containerization (CaaS)
• System Audits Services

• Data Lake Management

5
What Differentiates Us?

Intellectual Property
IP in Data Engineering
Framework cuts down the
end-to-end ML cycle by half

Focus on Innovation Experience Across Verticals


Contribution to Open-Source Highly Successful Engagements
Ecosystems, Adjudged 2020 Across the Spectrum, F100 Firms |
50 top technology company Top 3 Investment Banks | Silicon
Valley Firms

Highly Skilled Workforce Commitment To Data


All founders and 100% of We Live, Eat, Breathe Data,
workforce are from IIT 100% of our Business is
(Indian equivalent of MIT) about Data

6
Delivering Business Impact across the Data Analytics Value Chain

MTA and MMM 1:1 Personalized Marketing Churn Reduction Data Prediction and Correction

Used Quasi Experimental design to Built a multi-armed bandit solution Integrated 15+ data sources and
Built an ML model to fill gaps in
build a multi-touch attribution using reinforcement learning technique developed analytical datasets for modeling
product master data and correct
solution to improve the Effectiveness, to send personalized marketing to understand different customer
incorrect values with for 215000+
MMX, ROI and Sales of digital campaigns to existing customers to segments, improve customer loyalty and
products.
marketing campaigns increase the average ticket size per efficiency of targeted marketing
customers campaigns.

11% 2.5x 95%+


Increase in
7% Sales Uplift Increased accuracy of predicting Accuracy in Prediction
Marketing ROI the likelihood of churn

Automated Cloud data warehouse for Production-ise ML Models ETL and Data Warehousing
Data Ingestion improved analytics and
savings
Automated Data Ingestion from 10+ Developed an architecture to Processed huge volumes of customer
Retailers to enable real-time insights Developed a Spark-based ETL on GCP productionize the MAB model by and POS data, generating insights
into sales trends that enabled better to streamline and process 150 MN automating pipelines in AWS to within seconds for 500+ users
forecast demand and fulfil order rows of data ingested daily, resulting trigger personalized emails to through scalable and highly effective
management requests in 15x faster data availability and end customers data management using Hadoop and
lower maintenance costs. Spark

95% Reductions in
Time to Insights
2.5 MN Annual cost savings 24 MN+ Daily Orders
250 TB+ Annual
Data Volume

7
Full stack offerings to solve your data problems

Data Engineering Data Science & Analytics

Variety Volume Velocity Value

INGESTION ELT ANALYZE MODELLING VISUALIZATION


STRUCTURED DATA

EXTRACT LOAD TRANSFORM

UNSTRUCTURED DATA

DataOps

8
Data Engineering problems we solve for our clients

Data quality
Manual data Lack of data Business not able to
issues and lack
ingestion governance respond in real time
ELT of trust in data

High failure rate of Creating and Longer development


Model drift in Machine
deploying ML managing ML and deployment
Learning
MLOps models in production pipelines lifecycle

Data migration Lack of fault tolerant Non segregation of


Technology
from on-premise systems leading to Dev and test
selection for a
to cloud data data bad or environments from
CDW use case
warehouse incomplete data Production

9
Differentiated by data engineering capability and trained people

Open Source and Cloud-based


Data Engineering Expertise
Data Solutions

• Data engineering processes • Contributions to cloud and open-


adhered to solve big data problems source (Pig on Spark, First
Deployment of Spark on GCP)
• 100+ analytics models
operationalized • $50 MN Annual cloud spend
managed for clients

Data engineering
expertise
Outcome Focus Highly Skilled Talent Pool

• Problem refinement to articulate • Takshashila University to train


the use case employees in data and AI/ML

• Business focused from the • $3000 invested per employee to


start train as data engineer

10
Delivering value across the value chain

Front Middle Back

Revenue Product
Customer Sales & Operations & Manufacturing
Growth Development & Human Capital
Experience Marketing Supply Chain & Distribution
Management Innovation

Customer Sales Pricing Product Demand Predictive HR Analytics


Segmentation Analytics Analytics Innovation Planning & Maintenance
& Targeting Forecasting
Product Promotions,
Demand Affinity Product Production
Offers Inventory Planning Robotic
Monitoring Trends’ Optimisation
Optimisation & Out-of-Stock Process
Marketing Mix Analysis
Shopper Prevention Automation
Modelling Assortment Network
Behavior &
Planning & Sentiment Supply Chain
Trends Optimisation & Organisation
Multi-Touch Optimisation Analysis Planning Dynamic Routing Health
Customer Attribution
Loyalty & Churn Product Cost Analysis
Trade Spend Supply Chain
E-Commerce Optimisation Optimisation OEE &
Personalisation & Campaign Optimisation Performance
Hyper-Marketing Optimisation Analysis
Online
Merchandising
and Assortment

11
Data Science – Technologies and Platforms

BUSINESS USE CASES TECHNIQUES ALGORITHMS R&D

• Demand Forecasting • Classification and • Linear and logistic • Boosted Trees • AR, MA, • Reinforcement Learning
• Multidimensional Customer Clustering regression, ARIMA •
• Random Forest Quantum Computing
Segmentation • multinomial logistic • LDA
Graph Modeling • • Generative Adversarial
regression Logistic Regression
• Marketing Effectiveness • Knowledge • Apriori Networks (GAN)
• Latent class analysis • Decision Trees
• Customer Lifetime Value Representation • FP-Growth • Convolutional Neural
• Survival analysis • Neural Networks
Prediction • Time Series Modeling Networks (CNN)
• Bayesian
• Factor analysis • Markov Models
• Recommendation Systems • NLP • Recurrent Neural
• MAB
• Multivariate • CHAID Networks (RNN)
• Media Mix Modelling • Sensor Analytics
predictive modelling • Federated Learning
• Campaign Optimisation • Sequence Mining

TECH STACK

12
Takshashila: In-house learning academy to train data engineers

Challenges in training data engineers


• Most engineers' work on 10% development and 90% DevOps tasks
• Higher time spent on managing code that they have written over writing new ones
• Tied to 1 engineering stack, reluctance to try new technology as cost of failure is high
• The 3 Vs of data add an additional layer of risk beyond user acceptance risk, making projects inherently risky

Processes People
• Refined over 7 yrs to excel at execution • Cloud Architects
• Monitoring
• Project Scoping and Business alignment for • 10+ Data pipelines created yearly
• Deployment
early mitigation of user acceptance risk • Deep specialization in cloud
$10000+ infra training
• Data analytics for identification and cost • Cross cloud expertise
mitigation of data source & quality risk
Infrastructure
• Pluggable pipeline design using open
Technical Leads
source tools that remains agile with scale
• 15+ Data engineers managed
• Foundations in strong data engineering
Data • Cross functional training in analytics
fundamentals of high availability & Engineer
immutability & DevOps
• Ability to capture errors and roll back data
Statistics Algorithms • Manage execution risk
processing to
• Recover faster • Data Quality Developers
• Compression formats
• Reduce data processing cost • Probability • Significant training in cloud and opensource
• Consensus Protocol
• Failure rate • Production experience in building and
• Distributed algorithms
running TB Scale pipelines
• Data quality focus

13
Journey to become your Strategic Data to Outcomes Partner

Engagement Models

Capacity Project based Data Labs/ Build-Operate-


augmentation engagements Analytics CoE Transfer

Phase 1 Phase 2 Phase 3 Phase 4

• Exploratory calls to understand • First engagement kickoff • Enhance relationship by • Dedicated pool of resources
High

business, constraints, current • Showcase of capability with a engaging in more initiatives with varying level of experience
setup and technology primary engagement • First level understanding of data and capabilities
Business Impact

• 2-weeks consultation workshop • Alignment- understanding and means faster on-boarding, quick • Work ranging from everyday
to deep dive on the problems communications between teams solutioning and better sprints to long term projects
coordination

Continued Collaboration – Global Data Labs


Consultation Workshop and PoC Primary Engagement - MVPs Products
Low

0-3 Months 3-8 Months 8 - 12Months >12Months

14
Case study: Automated the MAB model to send 100M+ personalized emails
every 2 weeks for a global restaurant chain

Case Background
• There was a need to engage their customers through 1:1 Marketing
and in turn, maximize revenues, profits and CLTV 7%
• The new design had to be scalable and transferrable, with Sales Uplift
applicability across brands and geographies

Sigmoid Solution
14M
Customers
• Sanitised and refined the data to remove duplicity and redundancy Reached
and were also able to segment the whole customer data for the client
based on the features selected by the ML model
• Helped the client with KPI tracking and identifying offers which
worked better and were more profitable
100M
Mails per MAB
• Enabled the offers to be generated in a way so that they could be
personalised with subject line, creative content and offer type for
each customer
Tech Stack:

15
Case study: Personalized Recommendation boosts Profitability for a leading
CPG firm

Case Background
• The client is among the largest CPG firms in LATAM. They wanted to
develop a Personalized Digital Recommender System for their 24%
recently built website (which contributes to ~10% of overall sales) Improvement in
that would maximize average sales per order, conversion rate and Avg Sales Per Consultant
customer life-time value

Sigmoid Solution
8%
• Built an end to end system for generating personalized Improvement
in Profitability
recommendations (for digital subscribers) based on customers’
historical purchase history
• Implemented a multi-armed bandit based solution to learn
personalized recommendation strategy to select strategies
performing the best for a specific user at different time based on real-
100+
Strategies across
campaign feedback of consultants
8 countries
• Identified different profiles of consultants and developed
personalized strategies for each segment (Eg: cross-sell, up-sell,
Tech Stack: etc) to reduce initial strategy response learning period

16
Created schedule optimization dashboards with near real-time visibility on
operational functioning and different KPIs

The client a leading biopharmaceutical company with presence in over 100 countries

Case Background Sigmoid Solution Business Impact

• The client wanted to generate • The solution involved creating a Linear Programming model for the sites based on the
a production schedule using requirements & constraints. The model was created in python using ortools library and • Capacity Utilization
optimization techniques to optimises on capacity utilization.
determine, products that need
increased by 15%
to be manufactured, no. of Solution Highlights compared to manual
batches and the schedule scenario.
• Existing process was highly • ML models for gaining agility and efficiency in resource allocation
manual process for data • Automating data transmission and calculation process • Increased visibility
import/export, metric • Used Amazon SageMaker to build, train and deploy scalable models into real-time schedule
calculation and RCA. There • Developing schedule adherence visualization, notifications and trend highlights adherence tracking
was poor visibility into schedule
adherence
Traditional approaches follow a strict constraint inputs where this solution has been built • Faster actionable
• They wanted to automate this
keeping in mind to choose best constraints to optimize KPIs. The solution doesn’t include a
across multiple sites
single objective function but focuses on blended KPIs to achieve optimization
insights with deep-
considering maintenance dive analysis
windows, market demand and
number of manufacturing runs
per product – while catering to
specific requirements from
each site

17
Clients testify to our commitment and value creation

“The goals achieved by “Sigmoid’s solution helps us “Sigmoid has been a fantastic
Sigmoid’s data science and deliver best-in-class partner to Yum! Brands. They
data engineering teams have insights and services!” bring top talent with the
exceeded our expectations. best technical expertise to
They are exceptional in the table and consistently
understanding data and deliver great results. They
provide custom innovative are a trusted partner and a
solutions that directly impact pleasure to work with.”
the business revenue.”

Michael Christian R Paul Ryan Sabina Rizvi


Head of Data and Analytics Chief Technology Officer COO, YUM Digital & Technology

18
Thank You
Reimagine your business with data

Email: shaktip@sigmoidanalytics.com,
shakti.b@sigmoidanalytics.net

You might also like